Data center site selection and capacity optimization method based on improved sled dog algorithm

Abstract

The present application relates to the technical field of power system planning and data center layout optimization, and specifically discloses a data center site selection and capacity optimization method based on an improved snowshoe dog algorithm.The present application constructs a multi-objective optimization function with the minimum line loss, the minimum node voltage deviation and the minimum total cost as the target, and designs a collaborative solution framework based on the improved snowshoe dog algorithm, introduces a site selection tendency variable to generate a binary site selection scheme and a capacity mask repair strategy, and combines a penalty fitness screening mechanism of voltage safety constraints to solve the problems of slow convergence speed and easy falling into local optimum of traditional genetic algorithm and particle swarm optimization, and to overcome the defects of the original snowshoe dog algorithm lacking a hard constraint processing mechanism and the low efficiency of binary site selection repair, and to realize the comprehensive improvement of the solution speed and quality of the global optimal scheme of the collaborative planning of the distribution network and the computing center.

Description

Technical Field

[0001] This invention relates to the field of power system planning and data center layout optimization technology, and in particular to a collaborative optimization method for data center site selection and capacity determination based on an improved sled dog algorithm. Background Technology

[0002] In the construction of a new power system driven by the "dual carbon" goal, the distribution network needs to simultaneously undertake the dual tasks of power transmission and supporting the large-scale deployment of distributed computing centers. The joint optimization of "computing center location selection and capacity determination" has become the core issue of collaborative planning between the distribution network and the computing network. This problem is essentially a mixed-integer nonlinear programming problem, coupled with discrete location selection and continuous capacity configuration decisions. It also needs to meet multi-dimensional constraints such as voltage offset, power balance, and computing power scale, resulting in high solution complexity.

[0003] Existing traditional optimization algorithms, such as genetic algorithms and particle swarm optimization, generally suffer from slow convergence speed and susceptibility to local optima, making it difficult to output globally optimal solutions within engineering time constraints. While the Sled Dog Optimization (SDO) algorithm possesses strong global search capabilities, the original algorithm still has shortcomings when adapted to this problem: firstly, it lacks a hard constraint handling mechanism prioritizing feasibility, resulting in a large number of solutions that do not meet voltage offset constraints participating in the iteration, leading to low efficiency; secondly, traditional binary location methods rely on random updates, and continuous iteration lacks a clear direction, resulting in low efficiency in addressing structural constraints such as total computing power scale. Furthermore, existing research often focuses on single-objective optimization or fails to design algorithm improvements tailored to the coupling characteristics of the distribution network and computing center, making it difficult to simultaneously achieve optimal network losses, voltage quality, and cost. Therefore, an efficient optimization method adapted to this complex problem is urgently needed. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a data center location and capacity co-optimization method based on an improved sled dog algorithm.

[0005] In a first aspect, the present invention provides a data center location and capacity-determining collaborative optimization method based on an improved sled dog algorithm, the technical solution of which is as follows: Based on power system topology data, computing center construction data, and economic and renovation cost parameters, an objective function is constructed with the optimization objectives of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. Constraints are also constructed, including at least computing power demand constraints, node power balance constraints, node voltage safety constraints, and single-point construction scale constraints, to form a collaborative optimization model for data center site selection and capacity determination. The data center location and capacity co-optimization model is mapped to an optimization problem under the improved sled dog algorithm solution framework. The dimensions and encoding methods of the decision variables of the optimization problem are determined, and the parameters of the improved sled dog algorithm are initialized. The decision variables include location preference variables and capacity variables. For each individual in the improved sled dog algorithm population, a binary addressing variable is generated based on the addressing tendency variable. If all nodes are not selected, the node with the highest probability is forcibly selected as the addressing node. The capacity variable is masked based on the binary addressing variable, and the capacity variable of the unselected nodes is set to zero. The gap between the current total capacity and the total computing power requirement in the computing power requirement constraint is calculated. When the gap is positive, the gap is distributed among the selected nodes according to the proportion of the remaining capacity margin, and the capacity variable is updated. Power flow calculations are performed on individuals after decoding and constraint repair to obtain the voltage amplitude of each node. It is then determined whether the voltage amplitude of each node satisfies the node voltage safety constraint. For individuals that satisfy the node voltage safety constraint, a fitness value is calculated based on the objective function. For individuals that do not satisfy the node voltage safety constraint, a penalty fitness value is assigned. The penalty fitness value is set to be greater than the fitness values ​​of all individuals that satisfy the node voltage safety constraint. Based on the fitness value or the penalty fitness value, the individuals in the population are iteratively updated through the group cooperation mechanism of the improved sled dog algorithm until the iteration termination condition is met. Then, the location scheme and capacity configuration scheme corresponding to the best individual in the current population are output as the collaborative optimization result of data center location and capacity setting.

[0006] The beneficial effects of the data center site selection and capacity determination collaborative optimization method based on the improved sled dog algorithm of the present invention are as follows: The method of this invention constructs a multi-objective optimization function with the objectives of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. It designs a collaborative solution framework based on an improved sled dog algorithm, introduces a location preference variable to generate a binary location scheme and a capacity mask repair strategy, and combines a penalty fitness screening mechanism for voltage safety constraints. This solves the problems of slow convergence speed and easy getting trapped in local optima in traditional genetic algorithms and particle swarm algorithms. At the same time, it overcomes the shortcomings of the original sled dog algorithm, such as the lack of hard constraint processing mechanism and low efficiency of binary location repair. It achieves a comprehensive improvement in the speed and quality of solving the global optimal solution in collaborative planning between the distribution network and the computing center.

[0007] Based on the above scheme, the data center location and capacity co-optimization method based on the improved sled dog algorithm of the present invention can be further improved as follows.

[0008] In one alternative approach, the step of generating a binary addressing variable based on the addressing preference variable includes: The location preference variable is mapped to a location probability using the Sigmoid function, where the Sigmoid function is... ,in Indicates the first Individuals at candidate nodes Location preference variables, This represents the corresponding location probability; The location probability is determined according to a preset threshold. When the location probability is greater than the preset threshold, the binary location variable is set to 1; otherwise, it is set to 0.

[0009] The advantages of adopting the above-mentioned optional approach are as follows: by further mapping the location preference variable to the location probability through the Sigmoid function, and generating binary location variables based on a preset threshold, a smooth transformation from continuous variables to discrete location decisions is achieved, thereby improving the stability and interpretability of the location scheme generation.

[0010] In one alternative approach, the step of allocating the gap among the located nodes in proportion to the remaining capacity margin includes: Determine the set of selected nodes The set of selected nodes The set of node numbers for which the binary addressing variable takes a value of 1; Calculate the remaining capacity margin of each located node. ,in This represents the maximum capacity of a single node. According to the allocation formula Update the capacity variable, where For nodes Capacity variable, For the aforementioned gap, This represents the sum of the remaining capacity margins of the set of already located nodes.

[0011] The beneficial effects of adopting the above optional method are as follows: the capacity gap is further distributed among the selected nodes according to the proportion of the remaining capacity margin. By introducing the remaining capacity margin as the distribution weight, the capacity gap is reasonably allocated, ensuring the precise satisfaction of the total computing power demand constraint and optimizing the balance of capacity configuration.

[0012] In one alternative approach, the step of assigning a penalty fitness value to individuals that do not meet the node voltage safety constraints includes: Calculation of voltage feasibility markers If an individual satisfies the node voltage safety constraint, then ,otherwise ; According to the penalty fitness function Calculate the penalty fitness value, where The vector formed by the binary addressing variables, The vector formed by the capacity variables, This is a constant greater than the fitness value of all individuals that satisfy the node voltage safety constraints. and For preset coefficients, and This represents the voltage offset.

[0013] The beneficial effects of adopting the above-mentioned optional approach are as follows: by further introducing a penalty fitness value mechanism, individuals that do not meet the node voltage safety constraints are assigned a fitness value higher than that of all feasible solutions, which realizes the rapid identification and elimination of solutions that exceed the voltage limit, guides the population to evolve towards the voltage feasible region, and improves the strictness of constraint processing and search efficiency.

[0014] In one alternative approach, the voltage offset and The calculation method is as follows: , ,in This represents the minimum voltage amplitude at each node corresponding to the individual. This represents the maximum voltage amplitude at each node corresponding to the individual. The preset lower limit of voltage amplitude, This is the preset upper limit of voltage amplitude.

[0015] The advantages of adopting the above optional method are as follows: further using the maximum value function to calculate the voltage offset, accurately quantifying the degree of deviation of the node voltage amplitude from the safety boundary, providing a continuously differentiable metric for calculating the penalty fitness value, and enhancing the distinguishability of the degree of voltage constraint violation.

[0016] In one alternative approach, the group cooperation mechanism of the improved sled dog algorithm includes a sled dog selection mechanism and a sled dog movement mechanism; The sled dog selection mechanism determines the number of individuals participating in the sledding task based on their fitness values. The number of individuals participating in the sledding task ,in The population size is [not specified]. For values ​​in the interval random variables, For values ​​in the interval random variables; The sled dog movement mechanism is based on the individual's rank within the population and the number of individuals participating in the sled-pulling task. Different speed update formulas are used for the relationships between the dogs. The speed update of the lead dog is based on the group's optimal position and the individual's historical optimal position. The speed update of the middle dogs is based on the positions of their companions in front and behind. The speed update of the tail dog is based on the trajectory of the dog in front and introduces random perturbations.

[0017] The beneficial effects of adopting the above-mentioned optional methods are as follows: further design of sled dog selection and movement mechanisms, dynamic adjustment of the number of individuals participating in cooperation based on fitness ranking, and differentiated setting of speed update strategies for lead dogs, middle dogs and tail dogs, realizing adaptive regulation of population search behavior and balancing global exploration and local development capabilities.

[0018] In one alternative approach, the objective function is: ,in , , Preset weight parameters and satisfy The sub-objective of minimizing network loss is ,in For line resistance, and These are the active power and reactive power of the line, respectively. The node voltage amplitude is the minimum sub-target of the node voltage offset. ,in This is the rated voltage amplitude. Given the total number of nodes, the sub-objective with the minimum total cost is: ,in The fixed construction cost for a single node. Cost per unit length and unit capacity of line expansion For line length, For line flow, For line power flow limits, Cost of expanding substation capacity per unit area For the overall power flow of the system, This refers to the load limit for the substation.

[0019] The beneficial effects of adopting the above-mentioned optional approach are as follows: a three-objective weighted optimization function that includes line network loss, node voltage deviation and total cost is further constructed, which comprehensively considers the distribution network operation quality and the economic efficiency of computing center construction, realizes the synergistic trade-off of multi-dimensional optimization objectives, and provides a comprehensive evaluation framework for planning decisions.

[0020] Secondly, this invention provides a data center location and capacity optimization system based on an improved sled dog algorithm. The technical solution of this system is as follows: The module is used to construct an objective function based on power system topology data, computing center construction data, and economic and transformation cost parameters. The objective function is to minimize line network loss, node voltage deviation, and total cost. The module also constructs constraints including at least computing power demand constraints, node power balance constraints, node voltage security constraints, and single-point construction scale constraints, forming a collaborative optimization model for data center site selection and capacity determination. The determination module is used to map the data center site selection and capacity determination collaborative optimization model into an optimization problem under the improved sled dog algorithm solution framework, determine the dimension and encoding method of the decision variables of the optimization problem, and initialize the parameters of the improved sled dog algorithm. The decision variables include site selection bias variables and capacity variables. The update module is used to generate a binary addressing variable for each individual in the improved sled dog algorithm population based on the addressing tendency variable. If all nodes are not selected, the node with the highest probability is forcibly selected as the addressing node. The capacity variable is masked based on the binary addressing variable, and the capacity variable of the unselected nodes is set to zero. The gap between the current total capacity and the total computing power requirement in the computing power requirement constraint is calculated. When the gap is positive, the gap is distributed among the selected nodes according to the proportion of the remaining capacity margin, and the capacity variable is updated. The calculation module is used to perform power flow calculation on the individuals after decoding and constraint repair, obtain the voltage amplitude of each node, determine whether the voltage amplitude of each node meets the node voltage safety constraint, calculate the fitness value of individuals that meet the node voltage safety constraint according to the objective function, and assign a penalty fitness value to individuals that do not meet the node voltage safety constraint. The penalty fitness value is set to be greater than the fitness value of all individuals that meet the node voltage safety constraint. The optimization module is used to iteratively update the individuals in the population based on the fitness value or the penalty fitness value through the group cooperation mechanism of the improved sled dog algorithm until the iteration termination condition is met, and output the location scheme and capacity configuration scheme corresponding to the best individual in the current population as the collaborative optimization result of data center location and capacity.

[0021] The beneficial effects of the data center location and capacity co-optimization system based on the improved sled dog algorithm of the present invention are as follows: The system of this invention constructs a multi-objective optimization function with the goals of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. It designs a collaborative solution framework based on an improved sled dog algorithm, introduces a location preference variable to generate a binary location scheme and a capacity mask repair strategy, and combines a penalty fitness screening mechanism for voltage safety constraints. This solves the problems of slow convergence speed and easy getting trapped in local optima in traditional genetic algorithms and particle swarm algorithms. At the same time, it overcomes the shortcomings of the original sled dog algorithm, such as the lack of hard constraint processing mechanism and low efficiency of binary location repair. It achieves a comprehensive improvement in the speed and quality of solving the global optimal solution in collaborative planning between the distribution network and the computing center.

[0022] Thirdly, the technical solution of an electronic device according to the present invention is as follows: It includes a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps of the data center location and capacity co-optimization method based on the improved sled dog algorithm of the present invention.

[0023] Fourthly, the technical solution of a computer-readable storage medium provided by the present invention is as follows: The computer-readable storage medium stores instructions that, when read, cause the computer-readable storage medium to perform the steps of the data center location and capacity co-optimization method based on the improved sled dog algorithm of the present invention.

[0024] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0025] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart illustrating an embodiment of a data center location and capacity co-optimization method based on an improved sled dog algorithm according to the present invention. Figure 2 This is a schematic diagram of data center access. Figure 3 This is a schematic diagram of the SDO (Sled Dog Algorithm) process. Figure 4 This is a diagram illustrating the comparison of convergence speeds; Figure 5 This is a schematic diagram of the site selection and capacity distribution space. Figure 6 This is a schematic diagram of the voltage profile of the optimal solution; Figure 7 This is a schematic diagram of an embodiment of a data center location and capacity-determining collaborative optimization system based on an improved sled dog algorithm according to the present invention. Figure 8 This is a schematic diagram of an embodiment of an electronic device according to the present invention. Detailed Implementation

[0026] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0027] Figure 1 This diagram illustrates a flowchart of an embodiment of a data center site selection and capacity determination collaborative optimization method based on an improved sled dog algorithm provided by the present invention. This method can be executed by electronic devices such as terminal devices or servers. The terminal device can be any fixed or mobile terminal, such as a user equipment (UE), mobile device, user terminal, terminal, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, vehicle-mounted device, or wearable device. The server can be a single server or a server cluster consisting of multiple servers. Any electronic device can implement the data center site selection and capacity determination collaborative optimization method based on the improved sled dog algorithm by having its processor call computer-readable instructions stored in its memory. Figure 1 As shown, it includes the following steps: S100. Based on the power system topology data, computing center construction data, and economic and transformation cost parameters, construct an objective function with the optimization objectives of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. Construct constraints that include at least computing power demand constraints, node power balance constraints, node voltage safety constraints, and single-point construction scale constraints to form a collaborative optimization model for data center site selection and capacity determination.

[0028] Among them, power system topology data refers to the basic data set describing the physical architecture and electrical characteristics of the power grid, including the number, type, active load, reactive load, voltage amplitude, voltage phase angle, and upper and lower voltage limits of the bus or node; the location of the generator on the bus, its active output, reactive output, and upper and lower reactive limits; and the starting and ending bus, resistance, reactance, parallel susceptance, and capacity limit of the branch. For example, using a 33-node standard test system, the system contains 33 bus nodes, with node 1 being the balancing node and nodes 2 to 33 being the load nodes. The active load of each node is between 100kW and 200kW, and the safe operating range of the node voltage is 0.95pu to 1.05pu. The branch resistance and reactance parameters are set according to standard calculation examples.

[0029] Among them, the data for the construction of computing power centers refers to the planning parameters related to the computing power of data centers, including the total scale of computing power required for the region, the computing power-to-power conversion factor, the energy efficiency index, and the power factor; for example, the planned total computing power scale is 100 PFLOPS, the computing power-to-power conversion factor is 5 kW / PFLOPS, the energy efficiency index is 1.25, and the power factor is 0.95.

[0030] Among them, the economic and transformation cost parameters refer to the cost coefficients used to evaluate the economics of data center construction and power grid transformation, including the fixed construction cost of a single node, the cost of expanding a line per unit length and capacity, and the cost of expanding a substation per unit capacity; for example, the fixed construction cost of a single node is 500,000 yuan, the cost of expanding a line per unit length and capacity is 100,000 yuan / km / MW, and the cost of expanding a substation per unit capacity is 200,000 yuan / MVA.

[0031] Minimizing line network losses refers to optimizing the total active power loss of all transmission lines during power system operation. For example, in a 33-node system, optimizing the data center layout reduces line active power loss from 250kW to 188.66kW. Minimizing node voltage deviation refers to minimizing the sum of absolute deviations between the voltage amplitude of each node and its rated voltage. For example, in a 33-node system, optimization minimizes the sum of absolute deviations between the voltage amplitude of each node and 1.0pu. Minimizing total cost refers to reducing all economic inputs involved in data center construction and power grid upgrades, including fixed construction costs, line expansion costs, and substation expansion costs. For example, in a 33-node system, optimization reduces the total cost to 10.0238 million yuan.

[0032] In this context, the optimization objective refers to the evaluation criteria corresponding to the desired optimal state in a planning problem, used to measure the merits of different solutions. For example, in the data center site selection and capacity determination problem, the optimization objectives include minimizing network loss, minimizing node voltage deviation, and minimizing total cost. The objective function refers to the computational mathematical expression that transforms the optimization objective into a quantifiable assessment of the merits of each candidate solution. For example, the objective function F is equal to the weighted sum of the network loss sub-objective, the voltage deviation sub-objective, and the total cost sub-objective.

[0033] Constraints refer to the restrictive conditions that must be met in the planning problem to ensure the feasibility and security of the solution. For example, constraints include computing power requirement constraints, node power balance constraints, node voltage safety constraints, and single-node construction scale constraints. Computing power requirement constraints mean that the total computing power of all constructed data centers must reach or exceed the minimum total computing power required by the regional planning; for example, the total computing power of all selected nodes must be greater than or equal to 100 PFLOPS. Node power balance constraints mean that at each node, the injected active power and reactive power must equal the sum of the outflow power, node load, and data center load; for example, the input active power of node i minus the output active power equals the node's baseline active load plus the data center's active load plus active power losses. Node voltage safety constraints mean that the voltage amplitude of each node must be maintained within the range allowed for safe operation of the power grid; for example, the voltage amplitude of each node must be between 0.95 pu and 1.05 pu. The single-point construction scale constraint means that if each candidate node is selected to build a data center, its construction capacity must be between the minimum and maximum allowed capacity; for example, if node j is selected, its construction capacity must be between 0 PFLOPS and 24 PFLOPS.

[0034] Among them, the data center site selection and capacity determination collaborative optimization model refers to a mathematical programming model that aims to minimize network losses, node voltage deviations, and total costs while satisfying constraints such as computing power requirements, power balance, voltage safety, and single-point scale. For example, the collaborative optimization model established in a 33-node system aims to select several nodes from 32 candidate nodes and determine the construction capacity of each node, so as to achieve the minimum overall cost under the constraints of a total computing power of 100 PFLOPS and a voltage range of 0.95 pu to 1.05 pu.

[0035] S200. Map the data center location and capacity co-optimization model to an optimization problem under the improved sled dog algorithm solution framework, determine the dimension and encoding method of the decision variables of the optimization problem, and initialize the parameters of the improved sled dog algorithm. The decision variables include location preference variables and capacity variables.

[0036] Among them, the improved sled dog algorithm refers to a swarm intelligence optimization algorithm that improves the original sled dog optimization algorithm to address the characteristics of data center site selection and capacity determination problems. This includes introducing site selection bias variables and capacity mask repair strategies, as well as a penalty fitness screening mechanism for voltage hard constraints. For example, the improved sled dog algorithm changes the site selection variables from binary encoding to continuous bias variables, generates site selection decisions through Sigmoid mapping, and ensures the rapid satisfaction of computing power demand constraints through a capacity repair mechanism.

[0037] In this context, the optimization problem refers to transforming actual planning requirements into a solvable mathematical model within an algorithmic framework, encompassing decision variables, objective functions, and constraints. For example, the data center site selection and capacity allocation co-optimization problem, within the improved sled dog algorithm framework, is transformed into an optimization problem with 64-dimensional decision variables, three optimization objectives, and multiple constraints. Decision variables are those whose values ​​need to be determined in the optimization problem, and their values ​​determine a specific planning scheme. For instance, decision variables include 32 site selection bias variables and 32 capacity variables. The site selection bias variables determine whether each candidate node is selected, and the capacity variables determine the construction capacity of each selected node. Dimensions and encoding methods refer to the number of decision variables and their representation in the algorithm. For example, if the decision variables have 64 dimensions, the first 32 dimensions (site selection bias variables) are encoded using continuous real numbers, and the last 32 dimensions (capacity variables) are encoded using continuous real numbers.

[0038] The parameters of the improved sled dog algorithm refer to adjustable variables that control the algorithm's behavior and convergence performance, including population size, maximum number of iterations, search space boundary, and update strategy trigger threshold. For example, the population size is set to 50, the maximum number of iterations is set to 200, the search space boundary is from -10 to +10, the obstacle avoidance mechanism trigger threshold is set to 0.7, and the disorientation mechanism trigger threshold is set to 0.9. The location preference variable refers to a continuous variable used in the improved sled dog algorithm to characterize the degree to which a candidate node is selected. The location probability is obtained after mapping using the Sigmoid function. For example, if the location preference variable p-value for node j is 2.0, the location probability after Sigmoid mapping is approximately 0.88, indicating that node j has a high probability of being selected to build a data center. The capacity variable refers to a continuous variable used in the improved sled dog algorithm to characterize the construction capacity of each candidate node. After masking, the capacity of unselected nodes is set to zero. For example, the capacity variable s of node j is 15.0, which means that if node j is selected, the construction capacity will be 15.0 PFLOPS.

[0039] S300. For each individual in the improved sled dog algorithm population, generate a binary addressing variable based on the addressing tendency variable. If all nodes are not selected, force the node with the highest probability to be selected as the addressing node. Mask the capacity variable based on the binary addressing variable, set the capacity variable of the unselected nodes to zero, calculate the gap between the current total capacity and the total computing power requirement in the computing power requirement constraint, and when the gap is positive, distribute the gap among the selected nodes according to the proportion of the remaining capacity margin, and update the capacity variable.

[0040] Among them, the binary addressing variable refers to a 0-1 discrete variable generated by mapping the addressing preference variable. A value of 1 indicates that the corresponding node is selected to build a data center, and a value of 0 indicates that it is not selected. For example, the binary addressing variable of node 2 has a value of 1, which means that node 2 is selected to build a data center, and the binary addressing variable of node 3 has a value of 0, which means that node 3 is not selected.

[0041] Among these, the node with the highest probability is the one with the highest location probability after Sigmoid mapping among all candidate nodes. For example, when the binary location variables of all nodes are 0, node 2, with a location probability of 0.92, is forcibly selected as the location node. The selected node is the node whose binary location variable is 1, i.e., the node chosen to build the data center; for example, nodes 2, 19, 20, 21, and 22 are selected nodes.

[0042] Masking refers to using binary addressing variables to filter capacity variables, forcing the capacity variables of unselected nodes to be set to zero. For example, if node 3's binary addressing variable is 0, masking will change its capacity variable from 12.5 to 0, while node 2's binary addressing variable is 1, leaving its capacity variable 15.0 unchanged. Unselected nodes are those whose binary addressing variables are 0, meaning they were not selected for data center construction; for example, nodes 3, 4, and 5 are unselected nodes.

[0043] The current total capacity refers to the sum of the capacity variables of all selected nodes in the current planning scheme; for example, if node 2 has a capacity of 15.0 PFLOPS, node 19 has a capacity of 24.0 PFLOPS, node 20 has a capacity of 13.31 PFLOPS, node 21 has a capacity of 18.07 PFLOPS, and node 22 has a capacity of 20.62 PFLOPS, the current total capacity is 91.0 PFLOPS. The total computing power requirement refers to the minimum total computing power value required to be met by the planned area; for example, the total computing power requirement for the area is 100 PFLOPS.

[0044] The gap refers to the difference between the current total capacity and the total computing power demand. A positive difference indicates insufficient capacity that needs to be supplemented. For example, if the current total capacity is 91.0 PFLOPS and the total computing power demand is 100 PFLOPS, the difference is 9.0 PFLOPS, and the gap is positive 9.0 PFLOPS. The selected nodes are nodes whose binary addressing variable has a value of 1, meaning they have been selected to build a data center. For example, nodes 2, 19, 20, 21, and 22 are selected nodes. The remaining capacity margin ratio refers to the proportion of the capacity that each selected node can increase relative to the total increase in capacity of all selected nodes during the gap allocation process. For example, if node 2 has a remaining capacity margin of 5.0 PFLOPS and the total remaining capacity margin of all selected nodes is 20.0 PFLOPS, then node 2's remaining capacity margin ratio is 0.25.

[0045] S400. Perform power flow calculation on the individuals after decoding and constraint repair, obtain the voltage amplitude of each node, determine whether the voltage amplitude of each node meets the node voltage safety constraint, calculate the fitness value of individuals that meet the node voltage safety constraint according to the objective function, and assign a penalty fitness value to individuals that do not meet the node voltage safety constraint. The penalty fitness value is set to be greater than the fitness value of all individuals that meet the node voltage safety constraint.

[0046] The "individual after decoding and constraint repair" refers to the complete planning scheme that satisfies the computing power requirement constraints after the continuous variables in the algorithm are generated into binary addressing variables through Sigmoid mapping and threshold determination, and then processed by masking and capacity gap amortization. For example, after decoding and constraint repair, the individual results in a planning scheme with a construction capacity of 24.0 PFLOPS for node 2, 24.0 PFLOPS for node 19, 13.31 PFLOPS for node 20, 18.07 PFLOPS for node 21, and 20.62 PFLOPS for node 22.

[0047] Power flow calculation refers to a numerical calculation method that, under given power grid topology and load distribution conditions, determines the voltage magnitude and phase angle of each node, as well as the power distribution of each branch. For example, using the forward-backward substitution method to perform power flow calculation on a 33-node system, the voltage magnitude of node 2 is found to be 0.9674 pu, and the voltage magnitude of node 19 is found to be 0.9752 pu. The voltage magnitude of each node refers to the voltage magnitude of each bus node obtained through power flow calculation, expressed in per-unit value form; for example, after power flow calculation, the voltage magnitude of node 2 is found to be 0.9674 pu, the voltage magnitude of node 19 is 0.9752 pu, and the voltage magnitude of node 32 is 0.9501 pu.

[0048] Fitness value refers to a numerical value used to measure the quality of an individual; the smaller the value, the better the corresponding planning solution. For example, if one planning solution has a fitness value of 0.69 and another has a fitness value of 0.72, then the solution with fitness value 0.69 is better. Penalty fitness value refers to a very large value assigned to individuals that do not meet the node voltage safety constraint, ensuring that they are inferior to all individuals that meet the voltage safety constraint. For example, an individual that does not meet the voltage safety constraint is given a penalty fitness value of 10000, while the fitness values ​​of all individuals that meet the voltage safety constraint are less than 1. Therefore, infeasible solutions are at a disadvantage in population ranking.

[0049] S500. Based on the fitness value or the penalty fitness value, the individuals in the population are iteratively updated through the group cooperation mechanism of the improved sled dog algorithm until the iteration termination condition is met. Then, the location scheme and capacity configuration scheme corresponding to the best individual in the current population are output as the collaborative optimization result of data center location and capacity setting.

[0050] Among them, the group cooperation mechanism refers to the population update rules designed to simulate the cooperative behavior of sled dog groups in sledding tasks, including sled dog selection mechanism, movement mechanism, obstacle avoidance mechanism, disorientation mechanism, training mechanism and retirement mechanism; for example, individuals in the population are ranked according to their fitness values, the top 70% of individuals participate in sledding tasks, their positions are adjusted through a speed update formula, and some individuals perform obstacle avoidance operations or disorientation exploration operations.

[0051] Iterative updates refer to the process of continuously generating new generations of the population through a group cooperation mechanism during algorithm execution; for example, starting from the first generation population, a second generation population is generated through operations such as velocity updates, position updates, and boundary control. Iteration termination conditions refer to the conditions that stop algorithm iteration, including reaching the preset maximum number of iterations or the optimal fitness value no longer improving within a certain number of consecutive generations; for example, the algorithm terminates when the number of iterations reaches 200 generations, or when the optimal fitness value does not change within 50 consecutive generations.

[0052] Here, the current population refers to the set of all individuals in the current iteration round; for example, in the 100th generation iteration, the population contains 50 individuals, each corresponding to a site selection and capacity planning scheme. The optimal individual refers to the individual with the smallest fitness value in the current population, i.e., the best planning scheme found so far; for example, in the 200th generation population, the individual with a fitness value of 0.69 is the optimal individual, corresponding to the planning scheme of 24.0 PFLOPS for node 2, 24.0 PFLOPS for node 19, 13.31 PFLOPS for node 20, 18.07 PFLOPS for node 21, and 20.62 PFLOPS for node 22.

[0053] The site selection scheme refers to the decision of which nodes will be used to build the data center; for example, the site selection scheme might select nodes 2, 19, 20, 21, and 22 to be used to build the data center. The capacity configuration scheme refers to the decision of the capacity allocated to each selected node; for example, the capacity configuration scheme might be 24.0 PFLOPS for node 2, 24.0 PFLOPS for node 19, 13.31 PFLOPS for node 20, 18.07 PFLOPS for node 21, and 20.62 PFLOPS for node 22.

[0054] The collaborative optimization result refers to the final planning scheme that simultaneously meets the constraints of power grid operation safety and computing power demand, and achieves the optimal balance in terms of network loss, voltage quality and total cost. For example, the collaborative optimization result is to build data centers at nodes 2, 19, 20, 21 and 22, with a total capacity of 100 PFLOPS, a minimum voltage of 0.9674 pu, a total network loss of 188.66 kW and a total cost of 10.0238 million yuan.

[0055] The technical solution of this embodiment constructs a multi-objective optimization function with the goals of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. It designs a collaborative solution framework based on the improved sled dog algorithm, introduces a location preference variable to generate a binary location scheme and a capacity mask repair strategy, and combines a penalty fitness screening mechanism for voltage safety constraints. This solves the problems of slow convergence speed and easy getting trapped in local optima in traditional genetic algorithms and particle swarm algorithms. At the same time, it overcomes the defects of the original sled dog algorithm, such as the lack of hard constraint processing mechanism and low efficiency of binary location repair. It achieves a comprehensive improvement in the speed and quality of solving the global optimal solution in collaborative planning between the distribution network and the computing center.

[0056] In one alternative approach, the step of generating a binary addressing variable based on the addressing preference variable includes: The location preference variable is mapped to a location probability using the Sigmoid function, where the Sigmoid function is... ,in Indicates the first Individuals at candidate nodes Location preference variables, This represents the corresponding location probability.

[0057] The location probability refers to a value between 0 and 1 obtained by mapping the location propensity variable through the Sigmoid function, representing the likelihood that the corresponding node will be selected to build a data center. For example, the location propensity variable of node 2 is 2.0, and the location probability after Sigmoid mapping is 0.88. The location propensity variable of node 3 is -1.0, and the location probability after mapping is 0.27.

[0058] The location probability is determined according to a preset threshold. When the location probability is greater than the preset threshold, the binary location variable is set to 1; otherwise, it is set to 0.

[0059] The preset threshold refers to the boundary value used to convert the location probability into a binary location variable. For example, if the preset threshold is 0.5, nodes with a location probability greater than 0.5 are determined to be location nodes, and nodes with a probability less than or equal to 0.5 are not selected.

[0060] In the above-mentioned optional methods, the location preference variable is further mapped to the location probability through the Sigmoid function, and a binary location variable is generated according to a preset threshold. This realizes the smooth transformation of continuous variables into discrete location decisions and improves the stability and interpretability of the location scheme generation.

[0061] In one alternative approach, the step of allocating the gap among the located nodes in proportion to the remaining capacity margin includes: Determine the set of selected nodes The set of selected nodes This is the set of node numbers for which the binary addressing variable has a value of 1.

[0062] Calculate the remaining capacity margin of each located node. ,in This represents the maximum capacity of a single node.

[0063] According to the allocation formula Update the capacity variable, where For nodes Capacity variable, For the aforementioned gap, This represents the sum of the remaining capacity margins of the set of already located nodes.

[0064] Among the above-mentioned optional methods, the capacity gap is further distributed among the selected nodes according to the proportion of the remaining capacity margin. By introducing the remaining capacity margin as the distribution weight, the reasonable allocation of the capacity gap is achieved, ensuring the accurate satisfaction of the total computing power demand constraint and optimizing the balance of capacity configuration.

[0065] In one alternative approach, the step of assigning a penalty fitness value to individuals that do not meet the node voltage safety constraints includes: Calculation of voltage feasibility markers If an individual satisfies the node voltage safety constraint, then ,otherwise .

[0066] The voltage feasibility flag is a flag variable used to identify whether an individual meets the node voltage safety constraints. A value of 0 indicates that the voltage constraint is met, and a value of 1 indicates that the voltage constraint is violated. For example, the voltage amplitude range of individual A is 0.95pu to 1.03pu, and the voltage feasibility flag is 0. The voltage amplitude range of individual B is 0.93pu to 1.02pu, and the voltage feasibility flag is 1.

[0067] According to the penalty fitness function Calculate the penalty fitness value, where The vector formed by the binary addressing variables, The vector formed by the capacity variables, This is a constant greater than the fitness value of all individuals that satisfy the node voltage safety constraints. and For preset coefficients, and This represents the voltage offset.

[0068] Among the above-mentioned alternative approaches, a penalty fitness value mechanism is further introduced, which assigns a fitness value higher than that of all feasible solutions to individuals that do not meet the node voltage safety constraints. This enables the rapid identification and elimination of solutions that exceed the voltage limit, guides the population to evolve towards the voltage feasible region, and improves the strictness of constraint processing and search efficiency.

[0069] In one alternative approach, the voltage offset and The calculation method is as follows: , ,in This represents the minimum voltage amplitude at each node corresponding to the individual. This represents the maximum voltage amplitude at each node corresponding to the individual. The preset lower limit of voltage amplitude, This is the preset upper limit of voltage amplitude.

[0070] In the above-mentioned optional methods, the voltage offset is further calculated using the maximum value function, which accurately quantifies the degree of deviation of the node voltage amplitude from the safety boundary, providing a continuously differentiable metric for calculating the penalty fitness value and enhancing the distinguishability of the degree of voltage constraint violation.

[0071] In one alternative approach, the improved sled dog algorithm's group cooperation mechanism includes a sled dog selection mechanism and a sled dog movement mechanism.

[0072] The sled dog selection mechanism determines the number of individuals participating in the sledding task based on their fitness values. The number of individuals participating in the sledding task ,in The population size is [not specified]. For values ​​in the interval random variables, For values ​​in the interval A random variable.

[0073] The sled dog movement mechanism is based on the individual's rank within the population and the number of individuals participating in the sled-pulling task. Different speed update formulas are used for the relationships between the dogs. The speed update of the lead dog is based on the group's optimal position and the individual's historical optimal position. The speed update of the middle dogs is based on the positions of their companions in front and behind. The speed update of the tail dog is based on the trajectory of the dog in front and introduces random perturbations.

[0074] It should be noted that the improved sled dog algorithm's group cooperation mechanism also includes obstacle avoidance and disorientation mechanisms. When random numbers... When, perform obstacle avoidance maneuvers to update the individual's position; when When the individual is disoriented, perform a disorientation operation to update their location; when At that time, the individual's position remains unchanged, among which and For the preset threshold, For interval Random numbers within.

[0075] Among the above-mentioned optional methods, a sled dog selection mechanism and a movement mechanism were further designed. The number of individuals participating in the cooperation was dynamically adjusted according to the fitness ranking, and the speed update strategies of the lead dog, middle dog and tail dog were set differently. This achieved adaptive regulation of the population search behavior and balanced the global exploration and local development capabilities.

[0076] In one alternative approach, the objective function is: ,in , , Preset weight parameters and satisfy The sub-objective of minimizing network loss is ,in For line resistance, and These are the active power and reactive power of the line, respectively. The node voltage amplitude is the minimum sub-target of the node voltage offset. ,in This is the rated voltage amplitude. Given the total number of nodes, the sub-objective with the minimum total cost is: ,in The fixed construction cost for a single node. Cost per unit length and unit capacity of line expansion For line length, For line flow, For line power flow limits, Cost of expanding substation capacity per unit area For the overall power flow of the system, This refers to the load limit for the substation.

[0077] Among the above-mentioned optional approaches, a three-objective weighted optimization function is further constructed, which includes line network loss, node voltage deviation and total cost. This comprehensively considers the distribution network operation quality and the economic efficiency of computing center construction, and achieves a synergistic trade-off between multi-dimensional optimization objectives, providing a comprehensive evaluation framework for planning decisions.

[0078] In this embodiment, when a data center is connected to the power distribution network, a data center needs to be built at a selected node to form a "computing power-power" coupled system. Figure 2 The diagram shows a 33-node distribution network topology, where node 1 is the balancing node, and nodes 2 through 33 are load nodes. The data center can be connected to any of the nodes from node 2 to node 33.

[0079] During the model building phase, the power system topology data is first acquired, including bus or node numbers, types, active and reactive loads, voltage amplitude, voltage phase angle, and voltage upper and lower limits; generator bus locations, active and reactive power outputs, and reactive power upper and lower limits; and branch starting and ending buses, resistance, reactance, shunt susceptance, and capacity limits. Next, data on computing center construction is acquired, including the total computing power required for the region, computing power-to-power conversion coefficients, energy efficiency indicators, and power factor. Finally, economic and upgrade cost parameters are obtained, including the fixed construction cost of a single node, the cost of expanding a line per unit length and capacity, and the cost of expanding a substation per unit capacity.

[0080] Based on the above data, an objective function is constructed with the optimization objectives of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. The sub-objective of minimizing line network loss is: in For line resistance, and These are the active power and reactive power of the line, respectively. This represents the node voltage amplitude. The sub-target with the minimum node voltage offset is: in This is the rated voltage amplitude. Let be the total number of nodes. The sub-objective with the minimum total cost is: in The fixed construction cost for a single node. Cost per unit length and unit capacity of line expansion For line length, For line flow, For line power flow limits, Cost of expanding substation capacity per unit area For the overall power flow of the system, This represents the substation load limit. The overall objective function is: in , , Preset weight parameters and satisfy .

[0081] Simultaneously, constraints should be constructed that include at least computing power requirement constraints, node power balance constraints, node voltage security constraints, and single-point deployment scale constraints. The computing power requirement constraints are: in For binary addressing variables, The computing power of node i is determined by the node capacity. With energy efficiency indicators PUE and computing power power consumption conversion factor Power factor The calculation shows that: The node power balance constraint is: These represent the balance between active and reactive power, respectively. The node voltage safety constraints are: The voltage amplitude at all nodes must be maintained within the safe operating range. The single-point construction scale constraint is: in and These represent the minimum and maximum construction capacity of a single node, respectively. The objective function and constraints described above together constitute the data center site selection and capacity determination collaborative optimization model.

[0082] In the solution phase, the collaborative optimization model is mapped to an optimization problem within the framework of the improved sled dog algorithm. The dimensions and encoding methods of the decision variables in this optimization problem are determined, and the parameters of the improved sled dog algorithm are initialized. The decision variables include location preference variables and capacity variables, with a dimension of 2^n, where n is the number of candidate nodes. The first n dimensions represent the location preference variables, encoded using continuous real numbers, and the last n dimensions represent the capacity variables, also encoded using continuous real numbers. Initialized algorithm parameters include population size, maximum number of iterations, search space boundary, and update policy trigger threshold. Population initialization uses a random generation method, and the individual position vector is: in The lower boundary of the search space, The upper boundary, For interval Random numbers within, For population size, From a population perspective.

[0083] like Figure 3 As shown, the solution process of the improved sled dog algorithm begins with generating an initial population, followed by an iterative loop. In each iteration, the values ​​of all parameters are first updated, individuals are sorted according to an elimination mechanism, and the most suitable sled dogs are selected to participate in the sled pulling task. The sled dog selection mechanism determines the number of individuals participating in the sled pulling task based on their fitness values: in For population size, For values ​​in the interval random variables, For values ​​in the interval A random variable.

[0084] During the encoding / decoding and constraint repair phases, a binary addressing variable is generated for each individual in the population based on the addressing propensity variable. Specifically, the addressing propensity variable is mapped to an addressing probability using the sigmoid function. in Indicates the first Individuals at candidate nodes Location preference variables, This represents the corresponding location probability. The location probability is determined based on a preset threshold. If the location probability is greater than the preset threshold, the binary location variable is set to 1; otherwise, it is set to 0. If all nodes are not selected, the node with the highest probability is forcibly selected as the location node. The capacity variable is masked based on the binary location variable, and the capacity variable of unselected nodes is set to zero. The gap between the current total capacity and the total computing power requirement in the computing power demand constraint is calculated: in For the total capacity required by the plan, This represents the total capacity of the current plan. When the gap is positive, it is distributed among the selected nodes according to the proportion of remaining capacity margin. First, determine the set of selected nodes. Calculate the remaining capacity margin of each selected node. ,in This represents the maximum capacity of a single node. According to the allocation formula... Update the capacity variable, where This represents the total remaining capacity margin of the set of already located nodes.

[0085] In the fitness function calculation and constraint verification stage, power flow calculations are performed on the individuals after decoding and constraint repair to obtain the voltage amplitude of each node and determine whether the voltage amplitude of each node meets the node voltage safety constraints. To handle voltage constraints, a feasibility-first hard constraint mechanism is introduced. The original sled dog algorithm uses a soft penalty format: in For the addressing variable vector, For a vector of constant-volume variables, The original objective function is... The penalty coefficient is... To constrain the degree of violation, this method uses hard constraints, first defining voltage feasibility: in To preset the lower limit of voltage amplitude, The upper limit of the preset voltage amplitude is set. The final fitness function is: in This is a constant greater than the fitness value of all individuals that satisfy the node voltage safety constraints. and Voltage offset is a preset coefficient. In the population iteration and optimization mechanism stage, individuals in the population are iteratively updated based on fitness values ​​or penalized fitness values ​​through an improved group cooperation mechanism in the sled dog algorithm. This group cooperation mechanism includes sled dog selection, sled dog movement, obstacle avoidance, disorientation, training, and retirement mechanisms.

[0086] In the sled dog movement mechanism, the speed update formula for the lead dog is: The formula for updating the speed of mid-range dogs is: The formula for updating the speed of the dogs at the rear of the formation is: in For the individual's historical best position, The reference position is randomly selected from four optimal positions. Indicates the other dogs in the same group. To randomly generate individuals, The value can be 0 or 1. As the number of iterations decreases linearly from 1.0 to 0, , , For interval A random number within the range. Parameter and The calculation method is as follows: in This represents the current iteration number. The maximum number of iterations, Iterate from the following formula: initial value The position update formula is: In obstacle avoidance mechanisms, the formula for updating an individual's position is: in To take the value of integers, The individual with the worst fitness value, , For interval Random numbers within, and The calculation formula is: In the disorientation mechanism, the formula for updating an individual's position is: in To randomly generate individuals, , For interval Random numbers within, Let be a random variable that follows a standard normal distribution, and its probability density function be: To address the choice between obstacle avoidance and disorientation behaviors of the sled team, parameters are introduced. and If random numbers Perform obstacle avoidance maneuvers; if Perform a disorientation operation; if The position remains unchanged. For interval Random numbers within.

[0087] In the sled dog training mechanism, individuals who do not participate in sled pulling tasks are reassigned through specialized training: in , , , For interval Random numbers within, For learning ability: This represents the individual fitness value. (Gap) , , The calculation formula is: in To select the best individuals randomly from those performing the sledding task. For individuals not selected to perform the sledding task. Position update term during training. , , for: The sled dog retirement mechanism uses a random replacement method, with individuals ranked lower having a higher probability of retirement. In the boundary control strategy, when an individual exceeds the search boundary in any dimension, it is regenerated randomly. During the iteration termination and optimal solution output phase, the optimal fitness value of the population is recorded after each iteration to form a convergence curve, and the iteration process is continuously monitored. The iteration terminates when the number of iterations reaches the preset maximum number of iterations, or when the optimal fitness value stabilizes within a selected interval near the final optimal value. At this point, the position vector corresponding to the individual with the smallest fitness value in the population is the optimal solution. The final site selection scheme and the computing power capacity of each node are obtained through decoding.

[0088] like Figure 4As shown in the convergence speed comparison chart, the improved sled dog algorithm achieves a better fitness value and converges faster than the original sled dog algorithm, ant colony algorithm, and genetic algorithm with the same number of iterations. Figure 4 The horizontal axis represents the number of evolution steps, and the vertical axis represents the optimal fitness value. After 200 iterations, the improved sled dog algorithm achieves a fitness value of 0.69, while the original sled dog algorithm achieves 0.72, the ant colony algorithm achieves 0.14, and the genetic algorithm achieves 0.72. This indicates that the improved sled dog algorithm has higher solution accuracy and faster convergence speed.

[0089] like Figure 5 As shown, the calculation results of the site selection and capacity distribution show that the optimal construction method is to build a data center with a capacity of 24.0 PFLOPS at node 2, a data center with a capacity of 24.0 PFLOPS at node 19, a data center with a capacity of 13.31 PFLOPS at node 20, a data center with a capacity of 18.07 PFLOPS at node 21, and a data center with a capacity of 20.62 PFLOPS at node 22. Figure 5 The horizontal axis represents the selected node number, and the vertical axis represents the construction capacity, which intuitively displays the capacity configuration of each selected node.

[0090] like Figure 6 As shown, the optimal solution voltage profile indicates that under the optimal addressing and calibrating scheme, the voltage amplitude of each node is within the safe operating range, with the lowest voltage being 0.9674 pu, which meets the node voltage safety constraints. Figure 6 The horizontal axis represents the node number, and the vertical axis represents the voltage amplitude. The deviation of the voltage from the rated value is also marked, indicating that the voltage distribution is stable and no over-limit situations have occurred.

[0091] The final co-optimization result is a total computing power planning amount of 100 PFLOPS, a total network loss of 188.66 kW, and a total cost of 10.0238 million yuan. This embodiment achieves the co-optimization of line network loss, node voltage deviation, and total cost by using the above method, while meeting voltage safety constraints and computing power requirement constraints.

[0092] Figure 7 This diagram illustrates a structural schematic of an embodiment of a data center location and capacity co-optimization system 200 based on an improved sled dog algorithm provided by the present invention. Figure 7 As shown, the data center site selection and capacity determination collaborative optimization system 200 based on the improved sled dog algorithm includes: Module 201 is used to construct an objective function with the optimization objectives of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost based on power system topology data, computing center construction data, and economic and transformation cost parameters. It also constructs constraints including at least computing power demand constraints, node power balance constraints, node voltage security constraints, and single-point construction scale constraints to form a collaborative optimization model for data center site selection and capacity determination. The determination module 202 is used to map the data center location and capacity co-optimization model into an optimization problem under the improved sled dog algorithm solution framework, determine the dimension and encoding method of the decision variables of the optimization problem, and initialize the parameters of the improved sled dog algorithm. The decision variables include location preference variables and capacity variables. The update module 203 is used to generate a binary addressing variable for each individual in the improved sled dog algorithm population based on the addressing tendency variable. If all nodes are not selected, the node with the highest probability is forcibly selected as the addressing node. The capacity variable is masked based on the binary addressing variable, and the capacity variable of the unselected nodes is set to zero. The gap between the current total capacity and the total computing power requirement in the computing power requirement constraint is calculated. When the gap is positive, the gap is distributed among the selected nodes according to the proportion of the remaining capacity margin, and the capacity variable is updated. The calculation module 204 is used to perform power flow calculation on the individuals after decoding and constraint repair, obtain the voltage amplitude of each node, determine whether the voltage amplitude of each node meets the node voltage safety constraint, calculate the fitness value of individuals that meet the node voltage safety constraint according to the objective function, and assign a penalty fitness value to individuals that do not meet the node voltage safety constraint. The penalty fitness value is set to be greater than the fitness value of all individuals that meet the node voltage safety constraint. The optimization module 205 is used to iteratively update the individuals in the population based on the fitness value or the penalty fitness value through the group cooperation mechanism of the improved sled dog algorithm until the iteration termination condition is met, and output the location scheme and capacity configuration scheme corresponding to the best individual in the current population as the collaborative optimization result of data center location and capacity.

[0093] In an alternative embodiment, the update module 203 is specifically used for: The location preference variable is mapped to a location probability using the Sigmoid function, where the Sigmoid function is... ,in Indicates the first Individuals at candidate nodes Location preference variables, This represents the corresponding location probability; The location probability is determined according to a preset threshold. When the location probability is greater than the preset threshold, the binary location variable is set to 1; otherwise, it is set to 0.

[0094] In an alternative embodiment, the update module 203 is specifically used for: Determine the set of selected nodes The set of selected nodes The set of node numbers for which the binary addressing variable takes a value of 1; Calculate the remaining capacity margin of each located node. ,in This represents the maximum capacity of a single node. According to the allocation formula Update the capacity variable, where For nodes Capacity variable, For the aforementioned gap, This represents the sum of the remaining capacity margins of the set of already located nodes.

[0095] In an alternative embodiment, the computing module 204 is specifically used for: Calculation of voltage feasibility markers If an individual satisfies the node voltage safety constraint, then ,otherwise ; According to the penalty fitness function Calculate the penalty fitness value, where The vector formed by the binary addressing variables, The vector formed by the capacity variables, This is a constant greater than the fitness value of all individuals that satisfy the node voltage safety constraints. and For preset coefficients, and This represents the voltage offset.

[0096] In one alternative approach, the voltage offset and The calculation method is as follows: , ,in This represents the minimum voltage amplitude at each node corresponding to the individual. This represents the maximum voltage amplitude at each node corresponding to the individual. The preset lower limit of voltage amplitude, This is the preset upper limit of voltage amplitude.

[0097] In one alternative approach, the group cooperation mechanism of the improved sled dog algorithm includes a sled dog selection mechanism and a sled dog movement mechanism; The sled dog selection mechanism determines the number of individuals participating in the sledding task based on their fitness values. The number of individuals participating in the sledding task ,in The population size is [not specified]. For values ​​in the interval random variables, For values ​​in the interval random variables; The sled dog movement mechanism is based on the individual's rank within the population and the number of individuals participating in the sled-pulling task. Different speed update formulas are used for the relationships between the dogs. The speed update of the lead dog is based on the group's optimal position and the individual's historical optimal position. The speed update of the middle dogs is based on the positions of their companions in front and behind. The speed update of the tail dog is based on the trajectory of the dog in front and introduces random perturbations.

[0098] In one alternative approach, the objective function is: ,in , , Preset weight parameters and satisfy The sub-objective of minimizing network loss is ,in For line resistance, and These are the active power and reactive power of the line, respectively. The node voltage amplitude is the minimum sub-target of the node voltage offset. ,in This is the rated voltage amplitude. Given the total number of nodes, the sub-objective with the minimum total cost is: ,in The fixed construction cost for a single node. Cost per unit length and unit capacity of line expansion For line length, For line flow, For line power flow limits, Cost of expanding substation capacity per unit area For the overall power flow of the system, This refers to the load limit for the substation.

[0099] It should be noted that the beneficial effects of the data center site selection and capacity determination collaborative optimization system 200 based on the improved sled dog algorithm provided in the above embodiments are the same as those of the data center site selection and capacity determination collaborative optimization method based on the improved sled dog algorithm, and will not be repeated here. Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, and will not be repeated here.

[0100] The data center location and capacity determination collaborative optimization system 200 based on the improved sled dog algorithm of the present invention can be a computer program (including program code) running on a computer device. For example, the data center location and capacity determination collaborative optimization system 200 based on the improved sled dog algorithm of the present invention is an application software that can be used to execute the corresponding steps in the data center location and capacity determination collaborative optimization method based on the improved sled dog algorithm of the present invention.

[0101] In some embodiments, the data center location and capacity optimization system 200 based on the improved sled dog algorithm of the present invention can be implemented in a combination of hardware and software. As an example, the data center location and capacity optimization system 200 based on the improved sled dog algorithm of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the data center location and capacity optimization method based on the improved sled dog algorithm of the present invention. For example, the processor in the form of a hardware decoding processor can be one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0102] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0103] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned data center location and capacity co-optimization methods based on the improved sled dog algorithm. That is, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the data center location and capacity co-optimization method based on the improved sled dog algorithm shown in any embodiment of the present invention by calling the computer program.

[0104] In one alternative embodiment, an electronic device is provided, such as Figure 8 As shown, Figure 8 The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0105] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0106] Bus 4002 may include a path for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8The bus 4002 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0107] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0108] The memory 4003 stores application code (computer program) for executing the present invention, and its execution is controlled by the processor 4001. The processor 4001 executes the application code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0109] Among them, electronic devices can also be terminal devices. A terminal device can be any terminal device that can install applications and access web pages through applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0110] It should be noted that, Figure 8 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0111] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-mentioned data center location and capacity co-optimization methods based on the improved sled dog algorithm.

[0112] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0113] In an exemplary embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the aforementioned data center location and capacity co-optimization method based on the improved sled dog algorithm.

[0114] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed 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 remote computers, the remote computer can 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 can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0115] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0116] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0117] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0118] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0119] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0120] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0121] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A data center site selection and capacity determination collaborative optimization method based on an improved sled dog algorithm, characterized in that, include: Based on power system topology data, computing center construction data, and economic and renovation cost parameters, an objective function is constructed with the optimization objectives of minimizing line network loss, minimizing node voltage deviation, and minimizing total cost. Constraints are also constructed, including at least computing power demand constraints, node power balance constraints, node voltage safety constraints, and single-point construction scale constraints, to form a collaborative optimization model for data center site selection and capacity determination. The data center location and capacity co-optimization model is mapped to an optimization problem under the improved sled dog algorithm solution framework. The dimensions and encoding methods of the decision variables of the optimization problem are determined, and the parameters of the improved sled dog algorithm are initialized. The decision variables include location preference variables and capacity variables. For each individual in the improved sled dog algorithm population, a binary addressing variable is generated based on the addressing tendency variable. If all nodes are not selected, the node with the highest probability is forcibly selected as the addressing node. The capacity variable is masked based on the binary addressing variable, and the capacity variable of the unselected nodes is set to zero. The gap between the current total capacity and the total computing power requirement in the computing power requirement constraint is calculated. When the gap is positive, the gap is distributed among the selected nodes according to the proportion of the remaining capacity margin, and the capacity variable is updated. Power flow calculations are performed on individuals after decoding and constraint repair to obtain the voltage amplitude of each node. It is then determined whether the voltage amplitude of each node satisfies the node voltage safety constraint. For individuals that satisfy the node voltage safety constraint, a fitness value is calculated based on the objective function. For individuals that do not satisfy the node voltage safety constraint, a penalty fitness value is assigned. The penalty fitness value is set to be greater than the fitness values ​​of all individuals that satisfy the node voltage safety constraint. Based on the fitness value or the penalty fitness value, the individuals in the population are iteratively updated through the group cooperation mechanism of the improved sled dog algorithm until the iteration termination condition is met. Then, the location scheme and capacity configuration scheme corresponding to the best individual in the current population are output as the collaborative optimization result of data center location and capacity setting.

2. The data center site selection and capacity allocation collaborative optimization method based on the improved sled dog algorithm according to claim 1, characterized in that, The steps for generating binary addressing variables based on the addressing preference variables include: The location preference variable is mapped to a location probability using the Sigmoid function, where the Sigmoid function is... ,in Indicates the first Individuals at candidate nodes Location preference variables, This represents the corresponding location probability; The location probability is determined according to a preset threshold. When the location probability is greater than the preset threshold, the binary location variable is set to 1; otherwise, it is set to 0.

3. The data center site selection and capacity allocation collaborative optimization method based on the improved sled dog algorithm according to claim 2, characterized in that, The step of allocating the shortfall among the selected nodes according to the remaining capacity margin includes: Determine the set of selected nodes The set of selected nodes The set of node numbers for which the binary addressing variable takes a value of 1; Calculate the remaining capacity margin of each located node. ,in This represents the maximum capacity of a single node. According to the allocation formula Update the capacity variable, where For nodes Capacity variable, For the aforementioned gap, This represents the sum of the remaining capacity margins of the set of already located nodes.

4. The data center site selection and capacity optimization method based on the improved sled dog algorithm according to claim 1, characterized in that, The step of assigning a penalty fitness value to individuals that do not meet the node voltage safety constraints includes: Calculation of voltage feasibility markers If an individual satisfies the node voltage safety constraint, then ,otherwise ; According to the penalty fitness function Calculate the penalty fitness value, where The vector formed by the binary addressing variables, The vector formed by the capacity variables, This is a constant greater than the fitness value of all individuals that satisfy the node voltage safety constraints. and For preset coefficients, and This represents the voltage offset.

5. The data center site selection and capacity allocation collaborative optimization method based on the improved sled dog algorithm according to claim 4, characterized in that, The voltage offset and The calculation method is as follows: , ,in This represents the minimum voltage amplitude at each node corresponding to the individual. This represents the maximum voltage amplitude at each node corresponding to the individual. The preset lower limit of voltage amplitude, This is the preset upper limit of voltage amplitude.

6. The data center site selection and capacity allocation collaborative optimization method based on the improved sled dog algorithm according to claim 1, characterized in that, The improved sled dog algorithm's group cooperation mechanism includes a sled dog selection mechanism and a sled dog movement mechanism; The sled dog selection mechanism determines the number of individuals participating in the sledding task based on their fitness values. The number of individuals participating in the sledding task ,in The population size is [not specified]. For values ​​in the interval random variables, For values ​​in the interval random variables; The sled dog movement mechanism is based on the individual's rank within the population and the number of individuals participating in the sled-pulling task. Different speed update formulas are used for the relationships between the dogs. The speed update of the lead dog is based on the group's optimal position and the individual's historical optimal position. The speed update of the middle dogs is based on the positions of their companions in front and behind. The speed update of the tail dog is based on the trajectory of the dog in front and introduces random perturbations.

7. The data center site selection and capacity allocation collaborative optimization method based on the improved sled dog algorithm according to any one of claims 1 to 6, characterized in that, The objective function is: ,in , , Preset weight parameters and satisfy The sub-objective of minimizing network loss is ,in For line resistance, and These are the active power and reactive power of the line, respectively. The node voltage amplitude is the minimum sub-target of the node voltage offset. ,in This is the rated voltage amplitude. Given the total number of nodes, the sub-objective with the minimum total cost is: ,in The fixed construction cost for a single node. Cost per unit length and unit capacity of line expansion For line length, For line flow, For line power flow limits, Cost of expanding substation capacity per unit area For the overall power flow of the system, This refers to the load limit for the substation.

8. A data center site selection and capacity determination collaborative optimization system based on an improved sled dog algorithm, characterized in that, include: The module is used to construct an objective function based on power system topology data, computing center construction data, and economic and transformation cost parameters. The objective function is to minimize line network loss, node voltage deviation, and total cost. The module also constructs constraints including at least computing power demand constraints, node power balance constraints, node voltage security constraints, and single-point construction scale constraints, forming a collaborative optimization model for data center site selection and capacity determination. The determination module is used to map the data center site selection and capacity determination collaborative optimization model into an optimization problem under the improved sled dog algorithm solution framework, determine the dimension and encoding method of the decision variables of the optimization problem, and initialize the parameters of the improved sled dog algorithm. The decision variables include site selection bias variables and capacity variables. The update module is used to generate a binary addressing variable for each individual in the improved sled dog algorithm population based on the addressing tendency variable. If all nodes are not selected, the node with the highest probability is forcibly selected as the addressing node. The capacity variable is masked based on the binary addressing variable, and the capacity variable of the unselected nodes is set to zero. The gap between the current total capacity and the total computing power requirement in the computing power requirement constraint is calculated. When the gap is positive, the gap is distributed among the selected nodes according to the proportion of the remaining capacity margin, and the capacity variable is updated. The calculation module is used to perform power flow calculation on the individuals after decoding and constraint repair, obtain the voltage amplitude of each node, determine whether the voltage amplitude of each node meets the node voltage safety constraint, calculate the fitness value of individuals that meet the node voltage safety constraint according to the objective function, and assign a penalty fitness value to individuals that do not meet the node voltage safety constraint. The penalty fitness value is set to be greater than the fitness value of all individuals that meet the node voltage safety constraint. The optimization module is used to iteratively update the individuals in the population based on the fitness value or the penalty fitness value through the group cooperation mechanism of the improved sled dog algorithm until the iteration termination condition is met, and output the location scheme and capacity configuration scheme corresponding to the best individual in the current population as the collaborative optimization result of data center location and capacity.

9. An electronic device, characterized in that, The electronic device includes a processor coupled to a memory, the memory storing at least one computer program, which is loaded and executed by the processor to enable the electronic device to implement the data center site selection and capacity determination collaborative optimization method based on the improved sled dog algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer program, which, when executed by a processor, implements the data center location and capacity co-optimization method based on the improved sled dog algorithm as described in any one of claims 1 to 7.

Images