Green electricity direct connection method, system and device based on computing power elasticity mapping and medium

Abstract

The present application relates to the technical field of data center load flexible modeling and energy scheduling, and discloses a green electricity direct connection method, system, equipment and medium based on computing power elasticity mapping, energy side operation data, business side task data, thermal state side environment data and data center computing power load data are acquired to perform elastic task layering and time delay bucket processing, and obtain elastic load time delay bucket results; based on delay distribution variables and task conservation constraints, a feasible load boundary set is constructed through the bucket results, a temperature state recursive equation of the data center computer room is constructed in combination with the thermal state side environment data to modify the feasible load boundary set in the thermal state, obtain the total facility load boundary and take it as an input, solve the optimization model constructed based on the energy side operation data, and obtain the green electricity direct connection optimal operation decision; the business time elasticity which is difficult to be directly identified by the power side is converted into a standardized boundary input which can be directly called in the green electricity direct connection scene.

Description

Technical Field

[0001] This invention relates to the field of data center load flexibility modeling and energy scheduling technology, and in particular to a green electricity direct connection method, system, device and medium based on computing power elastic mapping. Background Technology

[0002] With the continuous development of technology, a large number of data centers have begun to adopt direct power supply from distributed photovoltaic and wind power. However, the output of new energy sources is significantly volatile and intermittent, while the load of traditional data centers exhibits continuous, stable, and rigid power consumption characteristics, resulting in a severe mismatch between the two in terms of time scale. How to transform the time elasticity of the data center's internal business side into identifiable and callable adjustment capabilities on the power side, while meeting the constraints of computing business completion and service quality constraints, and thus achieve more effective optimization of direct green power supply, has become a problem that needs to be solved.

[0003] In existing technologies, one type of solution starts from the constraints of the new energy side and performs flexible scheduling of computing tasks, such as patent CN121055283A. The technical focus of this type of solution is to form task scheduling results, which is essentially still an optimization arrangement at the business scheduling level. Another type of solution organizes the computing resource pool from the time dimension, such as patent CN113703936A. Although this type of solution involves task allowable windows, latency constraints and time rescheduling, its main purpose is usually to reduce the fragmentation of computing resources, improve resource scheduling efficiency, or directly incorporate deferred tasks as optimization variables into the scheduling model. Neither of these two types of solutions can accurately and stably transform data centers into flexible resources that can be standardized and called on the power side, and thus cannot achieve more effective optimization of green electricity direct connection. Summary of the Invention

[0004] This invention provides a method, system, device, and medium for direct green power connection based on computing power elastic mapping. It solves the problem of how to transform the time elasticity of the business side within the data center into an identifiable and callable adjustment capability of the power side while meeting the constraints of computing service completion and service quality constraints, thereby achieving more effective optimization of direct green power connection.

[0005] To address the aforementioned technical problems, the first aspect of this invention provides a method for direct green electricity connection based on computational power elastic mapping, comprising: Acquire energy-side operation data, business-side task data, thermal state-side environmental data, and data center computing load data for each time period within the scheduling cycle; Based on the maximum tolerable latency of the task, the business-side task data and the computing load data are processed by elastic task layering and latency bucketing to obtain elastic load latency bucketing results. Set delay allocation variables and combine them with task conservation constraints to construct the feasible load boundary set of the data center based on the elastic load delay bucketing results; Based on the feasible load boundary set and the thermal state side environment data, a temperature state recursive equation for the computer room in the data center is constructed to perform thermal state correction on the feasible load boundary set, thereby obtaining the total facility load boundary of the data center. An optimization model is established based on the energy-side operation data with the goal of minimizing the overall operating cost. The optimization model is then solved based on the load boundary and temperature constraints, using the total facility load boundary as input, to obtain the optimal operation decision for direct green electricity connection.

[0006] A second aspect of the present invention provides a green electricity direct connection system based on computing power elastic mapping, comprising: The data acquisition module is used to acquire energy-side operation data, business-side task data, thermal state-side environmental data, and data center computing load data for each time period within the scheduling cycle. The hierarchical and bucketing module is used to perform elastic task hierarchical and latency bucketing processing on the business-side task data and the computing load data according to the maximum tolerable latency of the task, so as to obtain the elastic load latency bucketing result. The boundary construction module is used to set delay allocation variables and, in conjunction with task conservation constraints, construct a set of feasible load boundaries for the data center based on the elastic load delay bucketing results. The thermal state correction module is used to construct a temperature state recursive equation for the computer room in the data center based on the feasible load boundary set and the thermal state side environmental data to perform thermal state correction on the feasible load boundary set and obtain the total facility load boundary of the data center. The optimization solution module is used to establish an optimization model with the goal of minimizing the overall operating cost based on the energy-side operation data, and to perform optimization solution on the optimization model based on the load boundary and temperature constraints, using the total facility load boundary as input, to obtain the optimal operation decision for direct green electricity connection.

[0007] A third aspect of the present invention provides an electronic device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the green electricity direct connection method based on computational power elastic mapping as described above.

[0008] A fourth aspect of the present invention provides a computer-readable storage medium comprising a stored computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the green electricity direct connection method based on computing power elastic mapping as described above.

[0009] Compared with the prior art, the beneficial effects of the embodiments of the present invention are as follows: By using latency binning, latency allocation variables, and task conservation, the latency tolerance of heterogeneous tasks is transformed into feasible IT-side load boundaries for each time period. This achieves a unified transformation from business-layer time elasticity to power-side boundary objects, making the data center's adjustability no longer a discrete scheduling result but a boundary range available for use in each time period. This allows the data center's flexibility to be called upon in a standardized manner by the power system. Incorporating temperature recursion equations into the boundary construction process avoids the "computing power-thermal dynamic mismatch" problem in the generated total facility load boundary, reducing overheating risk and cooling energy consumption. Using the thermally corrected total facility load boundary as input to the optimization model ensures that business-side constraints, thermal constraints, and energy-side constraints are unified and merged into boundary objects before entering optimization. This facilitates unified coupling with green electricity output, electricity price signals, and energy storage status, and improves the consistency between green electricity direct connection operation decisions and the actual executable capabilities of the data center. Under the premise of meeting computing service completion constraints and service quality constraints, the time elasticity of the data center's internal business side is transformed into identifiable and callable adjustment capabilities on the power side, thereby achieving more effective green electricity direct connection optimization. Attached Figure Description

[0010] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart of a green electricity direct connection method based on computing power elastic mapping provided in a certain embodiment of the present invention; Figure 2 This is a schematic diagram of the coupling between the equivalent load boundary, temperature state, and cooling power according to a certain embodiment of the present invention; Figure 3 This is a mapping diagram of the service side, energy side, and thermal state to load boundary provided in a certain embodiment of the present invention; Figure 4 This is a block diagram of green electricity direct connection optimization and dual feedback execution based on total facility load boundary and temperature constraints provided in a certain embodiment of the present invention; Figure 5 This is a structural diagram of a green electricity direct connection system based on computing power elastic mapping provided in a certain embodiment of the present invention; Figure 6 This is a structural diagram of an electronic device provided in a certain embodiment of the present invention; Figure label: Among them, 10 is the data acquisition module; 20 is the hierarchical and bucketing module; 30 is the boundary construction module; 40 is the thermal state correction module; 50 is the optimization solution module; 5000 is the electronic equipment; 5001 is the processor; 5002 is the bus; 5003 is the memory; and 5004 is the transceiver. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings and examples. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0013] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art will be able to understand the specific meaning of the above terms in this application according to the specific circumstances.

[0014] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is merely for describing specific embodiments and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0015] In one embodiment, such as Figure 1 As shown, the first aspect of the present invention provides a method for direct green electricity connection based on computational power elastic mapping, comprising: S1. Obtain energy-side operation data, business-side task data, thermal state-side environmental data, and data center computing load data for each time period within the scheduling cycle; wherein, the thermal state-side environmental data includes computer room temperature, outdoor temperature, and thermal inertia parameters; Specifically, on the energy side, the discrete time periods within a scheduling cycle (e.g., 96 time periods in 15-minute intervals) are obtained. tThe data includes the predicted local green electricity output, time-of-use electricity price, current state of charge of the energy storage system, rated capacity of the energy storage, maximum charging power and maximum discharging power of the energy storage, and the coefficient of performance of the cooling system, which are used as energy-side operation data. Among them, the predicted local green electricity output... P ren ( t It can be obtained by superimposing the predicted power of photovoltaic and wind power, that is: In the formula, P pv ( t ) indicates the first t Forecasted photovoltaic output for the specified time period; P wind ( t ) indicates the first t The predicted wind power output for the specified time period; it should be noted that these predicted data can be obtained by training a neural network model or by fitting historical data, which will not be elaborated on here.

[0016] On the business side, attribute information of the computational tasks to be executed is extracted from the task scheduling platform, container orchestration platform, or job queue management platform. For the first task within the scheduling cycle... i For each task, its arrival time, calculated demand, maximum tolerable delay, and business type are obtained as business-side task data, which are then used as input parameters for subsequent rigid / elastic discrimination and latency binning.

[0017] On the thermal state side, further acquire or set temperature-coupled state parameters during a certain scheduling cycle in the data center operation, including: t The outdoor ambient temperature, data center entrance temperature or server room temperature, lower temperature safety limit, upper temperature safety limit, and thermal inertia parameters, such as cooling system performance coefficient, data center equivalent heat capacity, and equivalent thermal resistance, are used as thermal state-side environmental data for a given time period. Among them, equivalent heat capacity and equivalent thermal resistance are used to characterize the overall thermal inertia characteristics of the data center server room, server cluster, and surrounding air environment at a given scheduling time period scale.

[0018] Simultaneously, the raw computing load data of the data center is acquired, and then this acquired data is uniformly formatted into structured data for discrete time periods within the scheduling cycle, eliminating differences between multi-source data. Through the unified collection of the above input information, the criteria for determining service time-shifting capabilities, energy regulation constraints, and basic parameters required for thermal state evolution are obtained simultaneously under the same scheduling framework. This provides basic data support for subsequent elastic task layering, latency binning, equivalent load boundary construction, and temperature coupling correction, and provides prerequisites for the subsequent mapping process from "service elasticity to power boundary" and the coupling process from "boundary to temperature-cooling correction".

[0019] S2. Based on the maximum tolerable latency of the task, perform elastic task layering and latency bucketing on the business-side task data and the computing load data to obtain elastic load latency bucketing results. In one embodiment, step S2 includes: By introducing a flexibility ratio, the computing load data is split into rigid load and elastic load, resulting in a load splitting result. Based on the maximum tolerable delay of the elastic task, the elastic load is divided into multiple delay buckets to obtain the elastic load power of each delay bucket in each time period, and combined with the load splitting result to form the elastic load delay bucketing result.

[0020] Specifically, based on the acquired business-side task data, this invention performs layered processing on the raw computing load data of the data center under baseline operating conditions, and buckets elastic tasks according to their maximum tolerable latency.

[0021] Assuming the data center is in the t The original total computing power load for the time period is P dc ( t This load can be broken down into a rigid load and an elastic load, that is: In the formula, P fix ( t A rigid load represents the power requirement for a task that must be completed in the present or a very short time and cannot be significantly delayed. P flex ( t ) represents the power demand for tasks that are allowed to be delayed within a certain time window.

[0022] To represent the adjustable proportions at different time periods, this invention introduces a flexibility ratio α(t), and the load splitting result is then: In the formula, α ( t The value range of ) is [0,1], and its size is determined by the proportion of delayable tasks in the task queue at the current time period.

[0023] Based on this, the present invention further divides the elastic load into multiple delay buckets according to the maximum tolerable delay for different tasks. Let the set of delay buckets be... B Each of the delay buckets b Corresponding to a maximum allowed delay in execution, then the first...t The total flexible load over a period of time can be expressed as: In the formula, P flex,b ( t ) indicates the first t The time period belongs to the delay bucket b The elastic load power.

[0024] Furthermore, for each task in the elastic load, its maximum tolerable delay di (in the number of time periods) is defined. A delay bucket set B = {1, 2, ..., D} is set, where D is the maximum allowable number of time periods (e.g., 24). Each elastic task is categorized into a delay bucket b = ceil(di / Δt), where Δt is the length of the time period. Thus, the elastic load power of delay bucket b within each time period t is equal to the sum of the power of all tasks belonging to bucket b arriving during that time period. This bucketing process discretizes the continuous or discrete delay tolerance into a finite number of time scales, facilitating subsequent modeling.

[0025] The rigid / elastic load splitting results, the elastic load bucket set, and the power values ​​of each bucket are combined to form the elastic load delay bucketing result. This invention, by introducing a flexibility ratio, clearly divides the original computing power load into rigid and elastic components, avoiding improper scheduling of non-delayable tasks and ensuring the real-time requirements of critical services. Further layering and bucketing the elastic load according to the maximum tolerable delay quantifies the adjustable time windows of different task groups, providing a more accurate feasible domain description for subsequent optimization models. The "delay tolerance" characteristic of the business layer is transformed into a modelable layered adjustment object on the power side, that is, splitting the mixed load demand into a rigid baseline plus multiple delay buckets of elastic components. This transforms the optimization decision variables from massive single-task scheduling to aggregated time-period power allocation, greatly reducing the dimensionality of variables and the complexity of the optimization problem, making it suitable for actual online rolling scheduling in data centers.

[0026] S3. Set delay allocation variables and, in conjunction with task conservation constraints, construct the feasible load boundary set of the data center based on the elastic load delay bucketing results; In one embodiment, step S3 includes: Based on the elastic load time delay binning results, the power of the elastic load corresponding to each time delay bin that is executed after a certain time delay in each time period is used as its corresponding delay allocation variable. The task conservation constraints are determined based on the delay allocation variables, and the feasible load boundary set of the data center in each time period is determined by the delay allocation variables and the task conservation constraints.

[0027] Specifically, this invention introduces a delay allocation variable for the elastic load in each delay bucket, quantitatively models the delayed execution process of tasks, and applies task conservation constraints to ensure that the total number of tasks remains unchanged and only time reordering occurs.

[0028] Based on the results of elastic load time delay binning, the power of the elastic load corresponding to each time delay bin, which is executed after a certain number of time delays in each time period, is used as its corresponding delay allocation variable. That is, for the elastic load belonging to time delay bin b, a delay allocation variable is defined. X b ( t , τ ), where τ represents the duration of the delayed execution, satisfying nonnegativity, that is: In the formula, X b ( t , τ ) indicates the first t Time period generated, belongs to the delay bucket b In the elastic load, it is delayed τ The power of execution in each time period. Since tasks are only allowed to be postponed by time and not advanced, this variable is always non-negative, and the postponement duration is... τ The delay should not exceed the maximum tolerable delay of the delay bucket. b .

[0029] To ensure that the original task load is not lost or duplicated during the time shift, this invention applies task conservation constraints to each time period and each delay bucket: This constraint means: in the first... t The time period belongs to the delay bucket b All elastic loads must be fully distributed within their allowed execution window; redistribution can only occur at the execution time, and the workload cannot be reduced or increased. To further characterize the determination of the data center within a time period based on various delay allocation variables and task conservation constraints under given conservation constraints. t The feasible load boundary set is as follows: In the formula, Ω( t To ensure that the tasks in each delay bucket satisfy conservation and delay constraints, the first... t The set of feasible load boundaries for equivalent power that may be generated by the time period data center (not a single scheduling result, but a boundary that can be called upon by the power side). t 0 represents any time period during which tasks are generated. This set is not a single scheduling result, but rather a range of load capacity determined by business elasticity and available for subsequent operational optimization.

[0030] This invention transforms the business attribute of "tasks can be executed with delay" into a mathematical feasible domain expression with strict constraints. This makes elastic tasks no longer just exist as an abstract scheduling concept, but become boundary objects that can directly participate in power optimization calculations. It accurately quantifies the elastic adjustable range of computing power, reduces the optimization difficulty, and adapts to various business priorities and time granularities.

[0031] S4. Based on the feasible load boundary set and the thermal state side environment data, construct the temperature state recursive equation of the computer room in the data center to perform thermal state correction on the feasible load boundary set, and obtain the total facility load boundary of the data center. In one embodiment, step S4 includes: The equivalent IT load of the data center in each of the time periods is determined based on the feasible load boundary set, and the equivalent heat generation power of the data center in each of the time periods is determined based on the equivalent IT load. The heat exchange rate is determined by the thermal inertia parameter and the equivalent heating power, and the heat dissipation is determined by the room temperature, the outdoor temperature, and the thermal inertia parameter. Based on the computer room temperature, the heat exchange rate, and the heat dissipation, the temperature state recursive equation is constructed, and the cooling coupling correction power of the cooling system in the data center is determined according to the temperature state recursive equation. The equivalent IT load is corrected by the cooling coupling correction power to obtain the total facility load boundary of the data center in each of the time periods.

[0032] Specifically, this invention uses a feasible load boundary set as a basis to reconstruct the arrival execution power of each delay bucket in each time period along the time axis, and combines them to obtain the total elastic arrival power. Then, rigid loads are superimposed to obtain the equivalent IT load of the data center in each time period. Since IT load operation will translate into heat accumulation inside the data center, the equivalent IT load and thermal inertia parameters of this invention determine the equivalent heat generation power of the data center in time period t as follows: In the formula, Q heat ( t The equivalent heating power is generated by the conversion of IT equivalent load. P eq ( t This represents the equivalent IT load. k h This is the conversion factor between equivalent IT load and equivalent heat generation.

[0033] This invention determines the heat exchange rate using thermal inertia parameters and equivalent heat generation power. It determines the heat dissipation based on the computer room temperature, outdoor temperature, and thermal inertia parameters. Furthermore, considering the thermal inertia of the computer room and server cluster, this invention combines these calculated data with the heat removed by the cooling system, based on the law of energy conservation: Change in heat within the computer room = Heat generated (Δt × ... Q heat ( t ) / C th Subtract the heat dissipation through the building envelope (Δt×( T in ( t )- T out ( t )) / ( C th × R th Subtract the heat removed by the cooling system (Δt×) COP ( t )× P cool ( t ) / C th Establish data center entrance temperature or server room temperature. T in ( t Temperature state recurrence relation: In the formula, T in ( t ) is the first t Data center entrance temperature or server room temperature during a given time period; T out ( t ) is the first t Outdoor ambient temperature during the time period; C th Equivalent heat capacity for data centers; R th The equivalent thermal resistance of the data center; P cool ( t ) is the first t The electrical power consumed by the cooling system during a given period; COP ( t ) for the cooling system in the first t The performance coefficient for a given time period. This equation characterizes the dynamic balance between heat generation within the data center, heat exchange with the external environment, and heat removal by the cooling system. It thus describes the temperature hysteresis caused by thermal inertia. A schematic diagram of the coupling between the equivalent load boundary, temperature state, and cooling power is shown below. Figure 2 As shown in the diagram, the mapping between the service side, energy side, and thermal state to the load boundary is as follows: Figure 3 As shown.

[0034] The cooling coupling correction power of the cooling system within the data center can be determined based on the temperature state recursive equation. Then, by correcting the equivalent IT load for each time period using the cooling coupling correction power, the total facility load boundary of the data center for each time period can be obtained. In the formula, P all ( t The total facility power demand is the sum of the IT equivalent load and the cooling system load.

[0035] This invention, based on a set of feasible load boundaries, further performs heat generation power conversion, temperature state recursion, and cooling power correction to ultimately generate the total facility load boundary (the core source of the feasible load boundary set → thermally corrected total boundary); by introducing a temperature state equation, P cool ( t No longer is P eq ( t Instead of being a simple proportional conversion term, it becomes a decision-making or state-correction quantity coupled with heat generation level, ambient temperature, thermal inertia, and cooling capacity. This completes the mapping process from "task latency tolerance—latency binning—latency allocation variable—arrival execution power—equivalent power load boundary—temperature state correction—total facility load." Through this process, the service time-shifting capabilities within the data center, which were originally difficult for the power system to directly identify, are reconstructed into operational boundary objects that consider both power boundaries and thermal state safety. The impact of thermal inertia on the load boundary is accurately accounted for, enabling joint optimization of the load required by IT tasks and the cooling load. This improves the safety and energy efficiency of temperature control and adapts to different climates and data center layouts.

[0036] In one embodiment, determining the equivalent IT load of the data center in each of the said time periods based on the feasible load boundary set includes: Based on the feasible load boundary set, the arrival execution power of each delay bucket in each time period is calculated according to each delay allocation variable; The total elastic arrival power of the data center in each of the time periods is determined based on the arrival execution power of each of the aforementioned periods; The equivalent IT load of the data center in each of the aforementioned rigid loads is obtained by superimposing each of the aforementioned total elastic arrival power.

[0037] Specifically, this invention reconstructs the actual execution status of tasks in each time period along a continuous time axis based on delay allocation variables and conservation constraints, forms the arrival execution power, and constructs the equivalent power load boundary of the data center on this basis.

[0038] First, based on the feasible load boundary set, reconstruct the power reaching execution along the time axis, for the delay bucket. b , define the first t Execution power during the time period R b ( t )for: In the formula, R b ( t ) for the first t The time period that actually arrives and is executed comes from the delay bucket. b The total power of tasks. This power may include tasks that are executed immediately in this time period, or tasks that were delayed from the previous time periods to be executed in the current time period.

[0039] By summing up the arrival power of all delay buckets, the total elastic arrival power for the current time period can be obtained. R ( t ): Adding the total elastic power to the rigid load yields the first... t Equivalent IT load of the data center during the time period: In the formula, P eq ( t After considering time-shifted reordering of business processes, the first... t The actual IT power requirements that the data center needs to handle during a given period, also known as the equivalent IT load. Compared to the original load curve, P eq ( t This already reflects the redistribution of business flexibility over time.

[0040] This invention aggregates the power of elastic loads that were originally generated and redistributed to different time periods by delaying the allocation of variables, thus obtaining the "arrival execution power" that is actually processed by data center IT equipment in each time period. This avoids the power misalignment caused by simply treating elastic loads as the loads of the original generation time periods, and ensures that the equivalent IT load can truly reflect the physical power demand after scheduling decisions. The seamless superposition of rigid loads and elastic loads forms a complete equivalent IT load, which not only ensures the real-time performance of critical services, but also makes full use of the adjustment capabilities of elastic loads. The structure is clear and facilitates subsequent thermal state correction and optimization modeling.

[0041] S5. Based on the energy-side operation data, establish an optimization model with the goal of minimizing the overall operating cost, and use the total facility load boundary as input to perform optimization solution based on load boundary and temperature constraints to obtain the optimal operation decision for direct green electricity connection. Based on the constructed total facility load boundary and temperature state equation, this invention combines green electricity output, electricity price signals, and energy storage status to establish an optimization model for direct green electricity operation of data centers. It solves for the power purchase, energy storage charging and discharging power, cooling power, and time-shifted execution results of flexible tasks in each discrete time period, thereby achieving coordinated optimization of economic operation, temperature-safe operation, and green electricity consumption of data centers.

[0042] Within the scheduling period T, this invention constructs an optimization objective function with the goal of minimizing the overall operating cost; wherein, the overall operating cost includes three parts: the cost of purchasing electricity from the external power grid, the cost of charging and discharging energy storage, and the cost of power curtailment penalty, and its mathematical expression is: In the formula, C total This represents the total operating cost within the scheduling period. c ( t ) is the first t Time-of-use electricity pricing for different time periods; P grid ( t ) is the first t The amount of electricity purchased from the external power grid during a given time period; P ch ( t )and P dis ( t ) are respectively the energy storage system in the first t Charging power and discharging power during a given time period; P cur ( t ) is the first t The amount of power abandoned during a given period; k bat The cost coefficient for energy storage charging and discharging;k cur Δ is the penalty coefficient for power curtailment. t The length of a single discrete time period.

[0043] The constraints of the optimization model include green electricity balance constraints, power balance constraints, energy storage SOC dynamic constraints, temperature safety constraints, IT-side boundary constraints constructed based on the delay allocation variables and the task conservation constraints, cooling power constraints, and non-negativity constraints on purchased power; among which, To characterize the green electricity utilization process, this invention defines the first... t Local green power output forecast for the period P ren ( t (Based on actual power utilization) P use ( t ) and abandoned power P cur ( t If the structure is as follows, then the green electricity balance constraint is: To ensure a balance between power supply and demand at any given time, power balance constraints are established for each node. t During the specified period, the sum of the power purchased from the external power grid, the actual power utilized by green electricity, and the power discharged by energy storage should equal the sum of the total data center facility load and the power charged by energy storage, i.e.: To ensure the safe and sustainable operation of energy storage systems, this invention establishes dynamic constraints on the evolution of the state of charge, also known as dynamic constraints on the state of charge (SOC) of energy storage: In the formula, SOC ( t ) is the first t State of charge of the energy storage system at the end of the period; η ch and η dis These are the energy storage charging efficiency and the discharging efficiency, respectively. E cap This refers to the rated capacity of the energy storage.

[0044] At the same time, the energy storage system must also meet the upper and lower safety boundary constraints: And meet the charging and discharging power limit constraints: In the formula, P ch,max and P dis,max These are the maximum charging power and maximum discharging power of the energy storage system, respectively. u ch ( t )and u dis ( t ) are 0-1 variables, representing the energy storage in the first and second moments respectively. t The period is either in a charging or discharging state to ensure that the energy storage system cannot be charged or discharged simultaneously at the same time.

[0045] To ensure that the temperature and cooling capacity are within a safe and feasible range, this invention introduces temperature safety constraints: In the formula, T min , T max These represent the minimum and maximum temperatures, respectively.

[0046] Apply upper and lower limit constraints to the cooling system power: In the formula, P cool,max This is the maximum allowable power for the cooling system. This constraint ensures that while pursuing economic efficiency and green energy consumption, the computer room temperature must still be kept within a safe operating range, thereby avoiding excessive localized temperature rise caused by solely pursuing power-side targets.

[0047] To ensure that the optimization results are consistent with the physical boundaries, this invention applies a non-negative constraint to the purchased power: There are also IT-side boundary constraints, which are a combination of delayed allocation variable definitions and conservation constraints. The decision variables of this optimization model include delayed allocation variables (non-negative continuous), cooling power (continuous), energy storage charging and discharging power (continuous), charging and discharging state binary variables (mutually exclusive), purchased power (continuous), actual utilized green power (continuous), and abandoned power.

[0048] Under the combined influence of the aforementioned objective function and constraints, this invention incorporates delay allocation variables, arrival execution power, and equivalent load boundaries into a unified optimization framework. A mixed-integer linear programming solver is used for global iterative calculations to ultimately obtain the optimal operation decision for direct green power connections. This includes the optimal power purchase, optimal energy storage charging and discharging power, optimal green power utilization, optimal power curtailment, cooling coupling correction power, total elastic arrival power, and the corresponding elastic task time-shift execution results for each time period within the scheduling cycle. Thus, the system completes the optimization process from "business elasticity boundary" to "green power direct connection operation decision considering temperature safety constraints."

[0049] In the optimization phase, this invention does not directly incorporate the original task variables into the green electricity direct connection optimization model. Instead, it uses the total facility load boundary, corrected for thermal conditions, as input. In this way, business-side constraints, thermal conditions constraints, and energy-side constraints are unified and merged into boundary objects before optimization. This facilitates unified coupling with green electricity output, electricity price signals, and energy storage status, and improves the consistency between green electricity direct connection operation decisions and the actual executability of the data center.

[0050] In one embodiment, after step S5, the following is included: The energy storage charging and discharging power of each of the above energy storage units is converted into active power control commands for the energy storage converter to control the execution on the energy storage side. The cooling coupling correction power is converted into cooling system control commands to control the cooling system to execute; Map the total elasticity arrival power to the number of active task instances to control the execution of the container orchestration platform; The real-time power purchase and real-time data center temperature after the instruction is executed are obtained to perform a dual feedback correction based on grid connection and temperature on the optimal operation decision of the green power direct connection, and thus obtain the corrected operation instruction.

[0051] Specifically, this invention maps the obtained green electricity direct connection optimization results into energy storage system control commands, computing power execution commands, and cooling system execution commands. It also forms a dual-feedback closed loop through actual grid-connected power monitoring and actual temperature monitoring, thereby achieving coordinated implementation of the optimization results at the physical and operational layers. The optimization results are sent to the energy storage control, computing power execution, and cooling execution layers, and a closed-loop execution process is performed using dual feedback correction based on actual grid-connected power and actual temperature. This is illustrated in the green electricity direct connection optimization and dual-feedback execution block diagram based on the total facility load boundary and temperature constraints, as shown below. Figure 4 As shown.

[0052] The optimal charging power obtained by the solution P ch ( t ) and optimal discharge power P dis (t This is directly converted into the active power control setpoint of the energy storage converter to control the energy storage device to perform charging or discharging operations within the corresponding time period. Correspondingly, it can also be based on the charging and discharging state variables. u ch ( t )and u dis ( t Switch the energy storage operation mode to ensure that the energy storage execution process is consistent with the power boundary in the optimization model.

[0053] The optimal cooling power obtained by the solution P cool ( t The data is mapped to cooling system control commands, including the start / stop status of cooling equipment, the set value of cooling power, and the operating settings of the air supply or liquid cooling system, so as to ensure that the actual temperature status tracks the target temperature trajectory corresponding to the optimization results as closely as possible.

[0054] The total elastic reach power obtained by the solution R ( t This can be further mapped to the number of active task instances in a container orchestration platform or task scheduling platform. The following mapping relationship can be used: In the formula, N task ( t ) is the first t The number of task instances that need to be activated and executed during the time period; k task This is a conversion factor for task instances per unit of power. Through this mapping, the "optimal execution power on the power side" can be transformed into the "specific execution scale on the computing power side," and the number of container instances enabled, the release rhythm of batch processing tasks, or the allocation ratio of computing resources can be controlled accordingly.

[0055] When further refinement of the energy consumption curve is needed, this invention can also be combined with dynamic voltage and frequency regulation technology to assist in adjusting the server's computing frequency, thereby improving the tracking accuracy of the actual IT load curve on the optimization results. It should be noted that in this step, the server and computing nodes remain online, and load changes are primarily achieved through adjustments to task execution rhythm and resource allocation ratios, rather than relying on frequent hard starts and stops for power regulation.

[0056] To improve the consistency between actual operation and optimization results, the system introduces an actual grid-connected power feedback mechanism. This mechanism acquires the real-time purchased power and real-time equipment room temperature after command execution, and defines the first [parameter / value] based on the real-time purchased power. t The grid-connected power deviation for the time period is: In the formula,P grid,act ( t ) is the first t The actual grid-connected power purchased during the time period was obtained from monitoring. P grid ( t The target power purchase is obtained by optimizing the solution.

[0057] Based on this grid-connected power deviation, subsequent power commands are corrected through feedback. This process can be represented as follows: In the formula, P cmd ( t () represents the revised execution power command; k e This is the feedback correction coefficient.

[0058] Furthermore, to improve the consistency between predicted thermal state and actual temperature changes, this invention introduces an actual temperature feedback mechanism. Based on the real-time data center temperature definition... t The temperature deviation over the period is: In the formula, T in,act ( t ) is the first t The actual monitored temperature of the computer room or entrance during the specified time period; T in ( t To optimize the solution of the corresponding temperature state obtained from the calculation.

[0059] Based on this temperature deviation, the present invention provides feedback correction for subsequent cooling power commands, which can be expressed as: In the formula, P cool,cmd ( t This is the revised cooling power command; k temp This is the temperature feedback correction factor.

[0060] The corrected instructions are combined to form corrected operating instructions. Through this feedback correction process, the present invention can dynamically fine-tune the cooling action in subsequent periods based on the deviation between the actual room temperature and the predicted temperature, thereby reducing the impact of prediction errors, execution errors, or equipment response lag on the overall operating effect.

[0061] This invention also maps the optimization results into computing power execution instructions, cooling execution instructions, and energy storage execution instructions, and combines grid-connected power feedback and temperature feedback for correction. This helps to reduce the impact of prediction errors and execution errors on the operating results, and improves the feasibility of the solution in real data centers. This closed-loop execution method further enhances the engineering applicability of the boundary and operating strategies. The method proposed in this invention can be deployed in data center energy management platforms or campus microgrid energy management systems. Moreover, this method has good compatibility and portability, and is suitable for the green transformation of newly built or existing data centers, with significant industrial application value.

[0062] This application addresses the problem of how to transform the time elasticity of the data center's internal business side into an identifiable and callable adjustment capability on the power side, while satisfying computing service completion constraints and service quality constraints, thereby achieving more effective green power direct connection optimization. It designs a green power direct connection method based on computing power elastic mapping. This method, through latency bucketing, latency allocation variables, and task conservation, transforms the latency tolerance of heterogeneous tasks into feasible IT-side load boundaries for each time period. This achieves a unified transformation from business-layer time elasticity to power-side boundary objects, making the data center's adjustability no longer a discrete scheduling result but a boundary range available for callable in each time period. This allows the data center's flexibility to be standardized and invoked by the power system. The method also incorporates temperature recursion... The process of incorporating the boundary construction process ensures that the generated total facility load boundary avoids the problem of "computing power-thermal dynamic mismatch," reducing the risk of overheating and cooling energy consumption. Using the thermally corrected total facility load boundary as the input to the optimization model ensures that business-side constraints, thermal state constraints, and energy-side constraints are unified and merged into boundary objects before entering the optimization process. This facilitates unified coupling with green electricity output, electricity price signals, and energy storage status, and improves the consistency between green electricity direct connection operation decisions and the actual executable capabilities of the data center. Under the premise of meeting the constraints of computing business completion and service quality, the time elasticity of the business side within the data center is transformed into an identifiable and callable adjustment capability of the power side, thereby achieving more effective green electricity direct connection optimization.

[0063] It should be noted that although the steps in the flowchart above are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order requirement for the execution of these steps, and they can be executed in other orders.

[0064] In another embodiment, such as Figure 5 As shown, a second aspect of the present invention provides a green electricity direct connection system based on computational power elastic mapping, comprising: The data acquisition module 10 is used to acquire energy-side operation data, business-side task data, thermal state-side environmental data, and data center computing load data for each time period within the scheduling cycle. The hierarchical and bucketing module 20 is used to perform elastic task hierarchical and latency bucketing processing on the business side task data and the computing load data according to the maximum tolerable latency of the task, and obtain the elastic load latency bucketing result. Boundary construction module 30 is used to set delay allocation variables and, in conjunction with task conservation constraints, construct the feasible load boundary set of the data center based on the elastic load delay bucketing results; The thermal state correction module 40 is used to construct a temperature state recursive equation for the computer room in the data center based on the feasible load boundary set and the thermal state side environmental data to perform thermal state correction on the feasible load boundary set and obtain the total facility load boundary of the data center. The optimization solution module 50 is used to establish an optimization model with the goal of minimizing the overall operating cost based on the energy-side operation data, and to perform optimization solution on the optimization model based on the load boundary and temperature constraints, using the total facility load boundary as input, to obtain the optimal operation decision for direct green electricity connection.

[0065] It should be noted that each module in the aforementioned green electricity direct connection system based on flexible computing power mapping can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module. For specific limitations regarding the green electricity direct connection system based on flexible computing power mapping, please refer to the limitations of the green electricity direct connection method based on flexible computing power mapping mentioned above; both have the same function and role, and will not be repeated here.

[0066] A third aspect of the present invention provides an electronic device comprising: Processor, memory, and bus; The bus is used to connect the processor and the memory; The memory is used to store operation instructions; The processor is configured to execute the operation instructions by calling the operation instructions, thereby causing the processor to perform the operation corresponding to the green electricity direct connection method based on computing power elastic mapping as shown in the first aspect of this application.

[0067] In one alternative embodiment, an electronic device is provided, such as Figure 6 As shown, Figure 6The illustrated electronic device 5000 includes a processor 5001 and a memory 5003. The processor 5001 and the memory 5003 are connected, for example, via a bus 5002. Optionally, the electronic device 5000 may also include a transceiver 5004. It should be noted that in practical applications, the transceiver 5004 is not limited to one type, and the structure of this electronic device 5000 does not constitute a limitation on the embodiments of this application.

[0068] Processor 5001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 5001 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.

[0069] Bus 5002 may include a path for transmitting information between the aforementioned components. Bus 5002 may be a PCI bus or an EISA bus, etc. Bus 5002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0070] The memory 5003 may be a ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM, CD-ROM 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 is not limited thereto.

[0071] The memory 5003 is used to store application code that executes the scheme of this application, and its execution is controlled by the processor 5001. The processor 5001 is used to execute the application code stored in the memory 5003 to implement the content shown in any of the foregoing method embodiments.

[0072] Among them, electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers.

[0073] The fourth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a green electricity direct connection method based on computational power elastic mapping as shown in the first aspect of the present application.

[0074] Another embodiment of this application provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the aforementioned method embodiments.

[0075] Furthermore, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0076] In summary, this invention relates to the field of data center load flexibility modeling and energy scheduling technology, and discloses a method, system, device, and medium for green power direct connection based on computing power elastic mapping. It acquires energy-side operational data, business-side task data, thermal state-side environmental data, and data center computing power load data to perform elastic task layering and latency binning, obtaining elastic load latency binning results. Based on latency allocation variables and task conservation constraints, a feasible load boundary set is constructed using the binning results. Combined with thermal state-side environmental data, a temperature state recursive equation for the data center's computer rooms is constructed to thermally correct the feasible load boundary set, obtaining the total facility load boundary. This boundary is then used as input to solve the optimization model constructed based on energy-side operational data, yielding the optimal operating decision for green power direct connection. This transforms the business time elasticity, which is difficult for the power side to directly identify, into a standardized boundary input that can be directly invoked in green power direct connection scenarios.

[0077] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0078] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for direct green electricity connection based on computational power elastic mapping, characterized in that, include: Acquire energy-side operation data, business-side task data, thermal state-side environmental data, and data center computing load data for each time period within the scheduling cycle; Based on the maximum tolerable latency of the task, the business-side task data and the computing load data are processed by elastic task layering and latency bucketing to obtain elastic load latency bucketing results. Set delay allocation variables and combine them with task conservation constraints to construct the feasible load boundary set of the data center based on the elastic load delay bucketing results; Based on the feasible load boundary set and the thermal state side environment data, a temperature state recursive equation for the computer room in the data center is constructed to perform thermal state correction on the feasible load boundary set, thereby obtaining the total facility load boundary of the data center. An optimization model is established based on the energy-side operation data with the goal of minimizing the overall operating cost. The optimization model is then solved based on the load boundary and temperature constraints, using the total facility load boundary as input, to obtain the optimal operation decision for direct green electricity connection.

2. The green electricity direct connection method based on computational power elastic mapping according to claim 1, characterized in that, The process of performing elastic task stratification and latency binning on the business-side task data and the computing load data based on the maximum tolerable latency of the task, to obtain elastic load latency binning results, includes: By introducing a flexibility ratio, the computing load data is split into rigid load and elastic load, resulting in a load splitting result. Based on the maximum tolerable delay of the elastic task, the elastic load is divided into multiple delay buckets to obtain the elastic load power of each delay bucket in each time period, and combined with the load splitting result to form the elastic load delay bucketing result.

3. The green electricity direct connection method based on computational power elastic mapping according to claim 2, characterized in that, The process of setting delay allocation variables and combining them with task conservation constraints to construct the feasible load boundary set of the data center based on the elastic load delay bucketing results includes: Based on the elastic load time delay binning results, the power of the elastic load corresponding to each time delay bin that is executed after a certain time delay in each time period is used as its corresponding delay allocation variable. The task conservation constraints are determined based on the delay allocation variables, and the feasible load boundary set of the data center in each time period is determined by the delay allocation variables and the task conservation constraints.

4. The green electricity direct connection method based on computational power elastic mapping according to claim 3, characterized in that, The thermal state-side environmental data includes the computer room temperature, outdoor temperature, and thermal inertia parameters; among which, Based on the feasible load boundary set and the thermal state-side environmental data, a temperature state recursive equation is constructed for the computer rooms within the data center to perform thermal state correction on the feasible load boundary set, thereby obtaining the total facility load boundary of the data center, including: The equivalent IT load of the data center in each of the time periods is determined based on the feasible load boundary set, and the equivalent heat generation power of the data center in each of the time periods is determined based on the equivalent IT load. The heat exchange rate is determined by the thermal inertia parameter and the equivalent heating power, and the heat dissipation is determined by the room temperature, the outdoor temperature, and the thermal inertia parameter. Based on the computer room temperature, the heat exchange rate, and the heat dissipation, the temperature state recursive equation is constructed, and the cooling coupling correction power of the cooling system in the data center is determined according to the temperature state recursive equation. The equivalent IT load is corrected by the cooling coupling correction power to obtain the total facility load boundary of the data center in each of the time periods.

5. The green electricity direct connection method based on computational power elastic mapping according to claim 4, characterized in that, Determining the equivalent IT load of the data center in each of the aforementioned time periods based on the feasible load boundary set includes: Based on the feasible load boundary set, the arrival execution power of each delay bucket in each time period is calculated according to each delay allocation variable; The total elastic arrival power of the data center in each of the time periods is determined based on the arrival execution power of each of the aforementioned periods; The equivalent IT load of the data center in each of the aforementioned rigid loads is obtained by superimposing each of the aforementioned total elastic arrival power.

6. The green electricity direct connection method based on computational power elastic mapping according to claim 4, characterized in that, The constraints of the optimization model include green electricity balance constraints, power balance constraints, energy storage SOC dynamic constraints, temperature safety constraints, IT side boundary constraints constructed based on the delay allocation variables and the task conservation constraints, cooling power constraints, and non-negative power purchase constraints; the optimal operation decision for green electricity direct connection includes the power purchase, energy storage charging and discharging power, cooling coupling correction power, and total elastic arrival power for each time period.

7. A method for direct green electricity connection based on computational power elastic mapping according to claim 6, characterized in that, The process of establishing an optimization model based on the energy-side operation data with the objective of minimizing overall operating costs, and then using the total facility load boundary as input to perform optimization on the model based on load boundary and temperature constraints to obtain the optimal operating decision for direct green power connection, includes: The energy storage charging and discharging power of each of the above energy storage units is converted into active power control commands for the energy storage converter to control the execution on the energy storage side. The cooling coupling correction power is converted into cooling system control commands to control the cooling system to execute; Map the total elasticity arrival power to the number of active task instances to control the execution of the container orchestration platform; The real-time power purchase and real-time data center temperature after the instruction is executed are obtained to perform a dual feedback correction based on grid connection and temperature on the optimal operation decision of the green power direct connection, and thus obtain the corrected operation instruction.

8. A green electricity direct connection system based on computational power elastic mapping, characterized in that, include: The data acquisition module is used to acquire energy-side operation data, business-side task data, thermal state-side environmental data, and data center computing load data for each time period within the scheduling cycle. The hierarchical and bucketing module is used to perform elastic task hierarchical and latency bucketing processing on the business-side task data and the computing load data according to the maximum tolerable latency of the task, so as to obtain the elastic load latency bucketing result. The boundary construction module is used to set delay allocation variables and, in conjunction with task conservation constraints, construct a set of feasible load boundaries for the data center based on the elastic load delay bucketing results. The thermal state correction module is used to construct a temperature state recursive equation for the computer room in the data center based on the feasible load boundary set and the thermal state side environmental data to perform thermal state correction on the feasible load boundary set and obtain the total facility load boundary of the data center. The optimization solution module is used to establish an optimization model with the goal of minimizing the overall operating cost based on the energy-side operation data, and to perform optimization solution on the optimization model based on the load boundary and temperature constraints, using the total facility load boundary as input, to obtain the optimal operation decision for direct green electricity connection.

9. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the green electricity direct connection method based on computing power elastic mapping as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the green electricity direct connection method based on computing power elastic mapping as described in any one of claims 1 to 7.

Images