A computing network collaborative planning method and device considering computing task timing scheduling

Abstract

The present invention relates to the field of computer technology, and in particular to a method and device for collaborative planning of computing network energy taking into account the timing scheduling of computing power tasks. The present invention considers the energy supply characteristics of the energy supply system and the flexible allocation of computing power tasks to plan the computing network system, considers the load demand of the computing network system to plan the energy supply system, and uses computing power tasks and energy supply as coupling points and boundary interactions for the computing network system planning and design and the energy supply system planning and design. Through multiple iterations, the optimal planning scheme for the integration of computing network energy is finally optimized, which can overcome the inconsistency problem of the configuration scheme of traditional independent planning, achieve a high degree of coupling between computing power tasks and renewable energy power generation, improve the renewable energy absorption rate, reduce the energy cost of data centers, realize multi-level interaction of computing power resources-computing power tasks-load demand-energy supply, achieve the optimal overall configuration scheme of computing network energy, and realize collaborative planning of computing network energy with the timing scheduling of computing power tasks.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and device for collaborative planning of computing networks that takes into account the timing scheduling of computing tasks. Background Art

[0002] Computing network energy planning includes two parts: computing power network system planning and energy supply system planning. Computing power network system planning is used to meet the needs of various computing power tasks and formulate the optimal computing power equipment construction plan. Energy supply system planning is used to meet the energy needs of the computing power system and formulate the optimal electricity, cooling, heat and other energy system supply plans to ensure that the computing power network system uses energy safely, economically and reliably.

[0003] Currently, computing network energy planning is typically performed separately from energy supply system planning. During the planning process, both the computing network system and the energy supply system are unknown. The allocation of computing tasks within the computing network system requires consideration of the energy supply characteristics of the energy supply system. Furthermore, energy supply system planning requires consideration of changes in the computing network system's load demand.

[0004] Due to the flexible and adjustable nature of computing tasks, batch workloads can be time-shifted. This allows for a high degree of coupling between computing tasks and renewable energy generation, based on energy pricing policies and renewable energy output curves. This improves renewable energy absorption and, while meeting computing task latency constraints, allows computing tasks to be shifted to periods with low electricity prices, reducing data center energy costs. The flexible allocation of computing tasks directly impacts the configuration of computing network system equipment. This also drives the load demand on the computing network system, indirectly impacting the energy supply system configuration. The energy supply system configuration, in turn, impacts the computing task allocation, which in turn influences the computing network system configuration.

[0005] The current problem of computing network energy planning is that computing power network system planning and energy supply system planning are usually carried out separately. Usually, the computing power network system plans and then proposes energy demand, and the energy supply system matches it, making it impossible for the two modules to interact and coordinate.

[0006] Based on this, there is an urgent need for a computing network collaborative planning method and device that takes into account the timing scheduling of computing power tasks to solve the technical problem of how to achieve computing network collaborative planning with the timing scheduling of computing power tasks. Summary of the Invention

[0007] In order to solve the technical problem of how to achieve computing network collaborative planning with computing task timing scheduling, an embodiment of the present invention provides a computing network collaborative planning method and device considering computing task timing scheduling.

[0008] In a first aspect, an embodiment of the present invention provides a method for collaborative planning of computing networks that considers the timing scheduling of computing tasks. The method is applied to a single-node computing network and includes:

[0009] Establishing a computing power network planning model and an energy supply planning model; wherein the computing power network planning model includes a first upper layer model and a first lower layer model, and the energy supply planning model includes a second upper layer model and a second lower layer model;

[0010] The first upper-layer model takes maximizing data center revenue and minimizing carbon emissions as objective functions, and takes latency constraints and computing power scale constraints as constraints, determines the center computing power scale, and sends the center computing power scale to the first lower-layer model;

[0011] The first lower-layer model takes maximizing data center revenue and minimizing carbon emissions as objective functions, and takes energy constraints and computing power scale constraints as constraints, determines the data center's carbon emissions and profits, and sends the data center's carbon emissions and profits to the first upper-layer model;

[0012] The first upper-layer model and the first lower-layer model are repeatedly iterated to determine a first optimal solution; wherein the first optimal solution includes a computing power equipment planning scheme and a computing power energy consumption load curve, and the computing power energy consumption load curve is used as an input of the energy supply planning model to perform coupled interaction;

[0013] The second upper-layer model takes maximizing profit and minimizing carbon emissions during the planning period as objective functions, and takes equipment planning capacity upper limit constraints and equipment commissioning constraints as constraints, determines equipment configuration strategies and planned capacity, and sends the equipment configuration strategies and planned capacity to the second lower-layer model;

[0014] The second lower-layer model uses minimizing the operating cost as the objective function, and uses power balance constraints, equipment operation constraints, power exchange constraints between the energy system and the upper network, and green power ratio constraints as constraints to determine the optimal operating plan and total operating cost, and sends the optimal operating plan and total operating cost to the second upper-layer model;

[0015] The second upper-layer model and the second lower-layer model are repeatedly iterated to determine a second optimal solution; wherein the second optimal solution includes an energy supply planning scheme, a power curve of power generation equipment, a public grid thermal power curve, and a green electricity trading curve, and the power curve of power generation equipment, the public grid thermal power curve, and the green electricity trading curve are used as inputs of the computing power network planning model for coupled interaction;

[0016] The computing power network planning model and the energy supply planning model are cyclically coupled and interacted to realize the computing network collaborative planning of computing power task timing scheduling.

[0017] In a second aspect, an embodiment of the present invention further provides a computing network collaborative planning device that considers computing task timing scheduling, including:

[0018] A model building module, configured to establish a computing power network planning model and an energy supply planning model; wherein the computing power network planning model includes a first upper layer model and a first lower layer model, and the energy supply planning model includes a second upper layer model and a second lower layer model;

[0019] A first data processing module is configured to determine the center computing power scale based on the objective function of maximizing data center revenue and minimizing carbon emissions, and based on the constraints of latency and computing power scale, and to send the center computing power scale to the first lower-layer model;

[0020] a second data processing module configured to determine the carbon emissions and profits of the data center using the first lower-layer model as objective functions and energy constraints and computing power scale constraints as constraints, and to send the carbon emissions and profits of the data center to the first upper-layer model;

[0021] a third data processing module configured to repeatedly iterate the first upper-layer model and the first lower-layer model to determine a first optimal solution; wherein the first optimal solution includes a computing power equipment planning scheme and a computing power energy consumption load curve, and the computing power energy consumption load curve is used as an input of an energy supply planning model for coupled interaction;

[0022] A fourth data processing module is configured for the second upper-layer model to determine the equipment configuration strategy and planned capacity with the maximum profit and the minimum carbon emission during the planning period as the objective function, and with the equipment planned capacity upper limit constraint and the equipment commissioning constraint as the constraint conditions, and to send the equipment configuration strategy and planned capacity to the second lower-layer model;

[0023] A fifth data processing module is configured to determine the optimal operation plan and total operation cost for the second lower-layer model using minimization of operation cost as the objective function, and using power balance constraints, equipment operation constraints, power exchange constraints between the energy system and the upper-level network, and green power ratio constraints as constraints, and to send the optimal operation plan and total operation cost to the second upper-layer model;

[0024] a sixth data processing module configured to iteratively perform the second upper-layer model and the second lower-layer model to determine a second optimal solution; wherein the second optimal solution includes an energy supply planning scheme, a power curve of power generation equipment, a public thermal power curve, and a green power trading curve, and the power curve of power generation equipment, the public thermal power curve, and the green power trading curve are used as inputs of a computing power network planning model for coupled interaction;

[0025] The cyclic coupling interaction module is used for cyclic coupling interaction between the computing power network planning model and the energy supply planning model, and has realized the computing power network collaborative planning of the computing power task timing scheduling.

[0026] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in any embodiment of the present invention is implemented.

[0027] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the method described in any embodiment of the present invention.

[0028] An embodiment of the present invention provides a method and device for collaborative planning of computing network energy taking into account the timing scheduling of computing power tasks. The present invention takes into account the energy supply characteristics of the energy supply system and the flexible allocation of computing power tasks to plan the computing network system, and takes into account the load demand of the computing network system to plan the energy supply system. The computing network system planning and design and the energy supply system planning and design use computing power tasks and energy supply as coupling points and boundary interactions. Through multiple iterations, the optimal integrated planning scheme of computing network energy is finally optimized. It can overcome the inconsistency problem of configuration schemes of traditional independent planning, achieve a high degree of coupling between computing power tasks and renewable energy power generation, improve the renewable energy absorption rate, reduce the energy consumption cost of data centers, and realize multi-level interaction of computing power resources-computing power tasks-load demand-energy supply, so as to achieve the optimal overall configuration scheme of computing network energy and realize collaborative planning of computing network energy with timing scheduling of computing power tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 This is a flow chart of a computing network collaborative planning method that considers computing task timing scheduling, provided by an embodiment of the present invention;

[0031] Figure 2 This is a hardware architecture diagram of an electronic device provided by an embodiment of the present invention;

[0032] Figure 3 This is a structural diagram of a computing network collaborative planning device that considers computing task timing scheduling, provided by an embodiment of the present invention;

[0033] Figure 4This is a diagram of a computing network collaborative planning architecture that takes into account the timing scheduling of computing tasks, provided by an embodiment of the present invention;

[0034] Figure 5 This is a structural diagram of a computing power network planning model provided by an embodiment of the present invention;

[0035] Figure 6 This is a structural diagram of an energy supply planning model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0037] Please refer to Figure 1 The embodiment of the present invention provides a method for collaborative planning of computing networks that considers the timing scheduling of computing tasks. The method is applied to a single-node computing network and includes:

[0038] Step 100: Establish a computing power network planning model and an energy supply planning model; wherein the computing power network planning model includes a first upper layer model and a first lower layer model, and the energy supply planning model includes a second upper layer model and a second lower layer model;

[0039] Step 102: The first upper-layer model uses maximizing data center revenue and minimizing carbon emissions as its objective function, and uses latency constraints and computing power scale constraints as constraints to determine the center computing power scale, and sends the center computing power scale to the first lower-layer model.

[0040] Step 104: The first lower-level model uses maximizing data center revenue and minimizing carbon emissions as objective functions, and uses energy constraints and computing power scale constraints as constraints to determine the data center's carbon emissions and profits, and sends the data center's carbon emissions and profits to the first upper-level model.

[0041] Step 106: The first upper-layer model and the first lower-layer model are repeatedly iterated to determine a first optimal solution; wherein the first optimal solution includes a computing power equipment planning scheme and a computing power energy consumption load curve, and the computing power energy consumption load curve is used as an input of the energy supply planning model for coupled interaction;

[0042] Step 108: The second upper-layer model uses the maximum profit and the minimum carbon emission during the planning period as the objective function, and the equipment planning capacity upper limit constraint and the equipment commissioning constraint as the constraint conditions to determine the equipment configuration strategy and planned capacity, and sends the equipment configuration strategy and planned capacity to the second lower-layer model;

[0043] Step 110: The second lower-layer model uses minimizing the operating cost as the objective function, and uses power balance constraints, equipment operation constraints, power exchange constraints between the energy system and the upper-level network, and green power ratio constraints as constraints to determine the optimal operating plan and total operating cost, and sends the optimal operating plan and total operating cost to the second upper-layer model.

[0044] Step 112: The second upper-layer model and the second lower-layer model are repeatedly iterated to determine a second optimal solution. The second optimal solution includes the energy supply planning scheme, the power curve of the power generation equipment, the public grid thermal power curve, and the green power trading curve. The power curve of the power generation equipment, the public grid thermal power curve, and the green power trading curve are used as inputs of the computing power network planning model for coupled interaction.

[0045] Step 114: The computing power network planning model and the energy supply planning model are cyclically coupled and interacted to realize the computing power network collaborative planning of the computing power task timing scheduling.

[0046] In an embodiment of the present invention, a computing power network planning model and an energy supply planning model are established; wherein, the computing power network planning model includes a first upper-layer model and a first lower-layer model, and the energy supply planning model includes a second upper-layer model and a second lower-layer model; the first upper-layer model takes the maximum revenue and minimum carbon emissions of the data center as the objective function, and takes the delay constraint and computing power scale constraint as the constraint conditions, determines the center computing power scale, and sends the center computing power scale to the first lower-layer model; the first lower-layer model takes the maximum revenue and minimum carbon emissions of the data center as the objective function, and takes the energy constraint and computing power scale constraint as the constraint conditions, determines the carbon emissions and profits of the data center, and sends the carbon emissions and profits of the data center to the first upper-layer model; the first upper model and the first lower model are repeatedly iterated to determine the first optimal solution; wherein, the first optimal solution includes a computing power equipment planning scheme (computing power equipment type, quantity + scale FLOPS and computing power task timing optimization curve) and a computing power energy consumption load curve, and uses the computing power energy consumption load curve as the input of the energy supply planning model for coupling interaction; the second upper-layer model Taking the maximum profit and minimum carbon emissions during the planning period as the objective function, and the upper limit constraint of equipment planning capacity and equipment commissioning constraint as the constraints, the equipment configuration strategy and planned capacity are determined, and the equipment configuration strategy and planned capacity are sent to the second lower-level model; the second lower-level model takes the minimum operating cost as the objective function, and takes the power balance constraint, equipment operation constraint, power exchange constraint between the energy system and the upper network, and green electricity proportion constraint as the constraints, to determine the optimal operating plan and total operating cost, and send the optimal operating plan and total operating cost to the second upper-level model; the second upper-level model and the second lower-level model are repeatedly iterated to determine the second optimal solution; among which, the second optimal solution includes the energy supply planning plan (energy system construction configuration type + capacity), power curve of power generation equipment, public grid thermal power curve and green electricity trading curve, and the power curve of power generation equipment, public grid thermal power curve and green electricity trading curve are used as inputs of the computing power network planning model for coupling interaction; the computing power network planning model and the energy supply planning model perform cyclic coupling interaction, and the computing network collaborative planning of computing power task timing scheduling has been realized.

[0047] Described below Figure 1 How to perform the steps shown.

[0048] Regarding step 102:

[0049] In one embodiment of the present invention, the objective function is to maximize data center revenue and minimize carbon emissions, including the following formula:

[0050] max f ntp

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071] S t,c =C t,c ·(E t,nom -E t,co2 ) / 10000

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078] f ep =f in -f c

[0079] f tp =f ep -f ep ·λ vat ·λ umt -f ep ·λ vat ·λ est

[0080] f ntp =(f ep -f ep ·λ vat ·λ umt -f ep ·λ vat ·λ est )·(1-λ int )

[0081] Where, The number of energy supply production equipment put into operation in year t; The capacity of the selected model of production equipment for the i-th energy supply; is the number of the i-th energy storage device put into operation in year t; is the capacity of the model selected for the i-th energy storage device; C inv is the initial investment cost; T is the life cycle; r is the capital discount rate; is the investment cost in year t; α lr,t is the loan ratio in year t; i is the number of energy supply production equipment; Ω1 is the set of energy supply production equipment; The unit capacity investment cost of the production equipment for the i-th energy supply; is the investment and construction capacity of the i-th energy supply production equipment in year t; j is the energy storage equipment number; Ω2 is the energy storage equipment set; is the unit capacity investment cost of the j-th energy storage device; is the investment and construction capacity of the j-th energy storage equipment in year t, The annual operation and maintenance cost per unit capacity of the production equipment supplying the i-th energy source; is the annual operation and maintenance cost per unit capacity of the j-th energy storage device, is the purchase cost of thermal power from the public grid in year t, is the medium- and long-term transaction cost of the i-th green electricity in year t, is the normal green electricity purchase cost in year t, is the penalty cost for the deviation of the i-th green power medium- and long-term transaction in the t-th year, is the natural gas purchase cost in year t, is the hydrogen purchase cost in year t, S is the number of typical daily scenarios per year; N s is the duration of the typical day scene s in a year; H is the total number of time periods in a day, C e,t,s,h P is the unit price of electricity purchased from the public grid during the h period of the typical day scenario in year t; e,t,s,h is the amount of electricity purchased from the public grid during the h period of the typical day scenario in year t, C le,t,i,s,h P is the purchase price of green electricity in the i-th medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; le,t,i,s,h C is the purchase amount of green electricity in the i-th medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year, ge,t,s,h P is the normal green electricity purchase price for the typical day scenario h period in year t; ge,t,s,h is the normal green electricity purchase amount in the h period of the typical day scenario in year t, C g,t,s,h G is the natural gas purchase price in the typical day scenario h period in year t; g,t,s,h is the natural gas purchase amount in the typical day scenario h period in year t, The unit price of hydrogen purchased during the h period of the typical day scenario in year t is: is the amount of hydrogen purchased in the typical day scenario h period in year t, C le,t,i,s,h,dp The penalty price for deviation of green electricity medium- and long-term transaction in the i-th green electricity medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; P le,t,i,s,h,dp P is the deviation of the i-th green electricity medium- and long-term transaction caused by the responsibility of the power purchaser in the h-th period of the typical day scenario in the t-th year; lest,t,i,s,h The transaction volume of the i-th green power medium- and long-term trading contract in the h-th period of the typical day scenario in the t-th year; P legm,t,i,s,h The amount of green electricity provided by medium- and long-term transactions in the i-th period of time in the typical day scenario of the t-th year is C fc,t is the financial cost in year t; r i,tt is the annual loan interest rate for the loan in year tt; C inv,tt is the investment cost in year tt; α lr,tt is the loan ratio in year tt; N tt is the repayment period of the loan in year tt, f c is the total investment cost of the integrated energy system; f t,c is the investment cost in year t, SR t is the energy supply income in year t, C e,t,s,h,S are the electricity selling price in the typical day scenario h period in year t; P eL,t,s,h P is the other electric load in the typical day scenario h period in year t; cpeL,t,s,h HPR is the power load consumed by computing equipment in the typical day scenario h period in year t; FJis the heating price per unit area; CPR FJ is the cooling price per unit area; A r,HFJ is the heating area; A r,CFJ is the cooling area, S t,up is the surplus access income in year t; S is the number of typical daily scenarios each year; N s is the duration of the typical day scene s in a year; H is the total number of time periods in a day, C up,e,t,s,h P is the unit price of surplus electricity on the grid in the h period of the typical day scenario in year t; up,e,t,s,h is the on-grid power consumption in the h period of the typical day scenario in year t, formula S inv Subsidize total investment; is the investment subsidy for year t; i is the number of the energy supply production equipment; Ω1 is the set of energy supply production equipment; Subsidy for initial investment per unit capacity of the i-th energy supply production equipment; is the investment and construction capacity of the i-th energy supply production equipment in year t; j is the energy storage equipment number; Ω2 is the energy storage equipment set; is the initial investment subsidy per unit capacity of the jth energy storage device; is the investment capacity of the j-th energy storage equipment in year t, S t,sub is the power generation subsidy income in year t; S is the number of typical day scenarios each year; N s is the duration of the typical day scene s in a year; H is the total number of time periods in a day, C wt,e,t,s,h 、C pv,q,t,s,h are the electricity subsidies for wind power generation and photovoltaic power generation in the h period of the typical day scenario in year t; P wt,e,t,s,h 、P pv,e,t,s,h are the wind power generation and photovoltaic power generation in the typical day scenario h period in year t, S t,c is the environmental benefit in year t; C t,c is the carbon trading price in year t; E t,nom is the carbon quota for year t; E t,co2 is the carbon dioxide emissions in year t; α e,CO2 , is the emission coefficient of carbon dioxide generated by the consumption of electricity; α g, i,CO2 is the emission coefficient of carbon dioxide produced by the i-th energy supply production equipment consuming natural gas, S t,i,le,dp is the deviation income of the i-th green power medium- and long-term transaction in the t-th year, C le,t,i,s,h,dp The penalty price for deviation of green electricity medium- and long-term transaction in the i-th green electricity medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; P le,t,i,s,h,dp P is the deviation of the i-th green electricity medium- and long-term transaction caused by the responsibility of the power purchaser in the h-th period of the typical day scenario in the t-th year; lest,t,i,s,h The transaction volume of the i-th green power medium- and long-term trading contract in the h-th period of the typical day scenario in the t-th year; Plegm,t,i,s,h The amount of green electricity provided by medium- and long-term transactions in the i-th period of time in the typical day scenario of the t-th year is E RV is the total residual value of the integrated energy system equipment at the end of the planning period, M n is the total number of equipment in the integrated energy system, δ i is the residual value rate of the i-th equipment, C INV,i is the initial investment of the i-th equipment, f in is the total benefit of the integrated energy system over its entire life cycle; f t,in is the total income in year t, f ep For comprehensive energy services full life cycle economic benefits, f ntp is the total net profit over the entire life cycle, λ vat is the value-added tax rate, λ umt Maintaining tax rates for urban construction, λ est is the education surcharge rate, which is 5%, λ int is the income tax rate, f ntp is the total net profit over the entire life cycle, f e is the carbon dioxide emissions, T is the planning period, i.e. the total planning period; is the CO2 emission in year t; S is the number of typical daily scenes per year; N s is the duration of the typical day scene s in a year; H is the total number of time periods in a day; The carbon dioxide emission coefficient generated by consuming public grid electricity; is the carbon dioxide emission coefficient produced by natural gas.

[0082] In one embodiment of the present invention, the computing power task demand model is:

[0083] Meaning: Time within a day can be expressed as The task arriving at time slot t can be expressed as:

[0084]

[0085] Where:

[0086] As can be seen from 3.2.3 Task Feature Model;

[0087] x∈{delay-tolerant, delay-sensitive}, indicating the type of tasks classified according to their sensitivity to delay;

[0088] sce∈{working day, holiday, weekend}, represents a typical day divided by the amount of computation;

[0089] I represents the number of routers, i = [1, I] represents the router number;

[0090] Indicates a task The corresponding computing power resource requirements;

[0091] Indicates a task The corresponding network resource requirements;

[0092] Indicates a task The corresponding energy resource demand;

[0093] N i，x represents the number of tasks of type x that arrive at router i, n=[1,N i，x ] indicates the number of a specific task;

[0094] t∈{1,2,...,24}, represents the arrival time slot.

[0095] In a single data center, there is no cross-data center scheduling of tasks, so it is task-oriented.

[0096] Task, only time decision variable, determines the task The scheduling decision for forwarding time is

[0097] Delay Constraints

[0098]

[0099] Meaning: The total latency of each task is defined as the sum of transmission delay, propagation delay and processing delay.

[0100] The delay under scheduling decision Z can be expressed as:

[0101]

[0102] Where: is the transmission delay, that is, the task is summed up through the servers on the transmission path, is the propagation delay (actual physical distance divided by the speed of light), and is the computational delay. The computational delay of each task can be expressed as:

[0103]

[0104] Where: Indicates the computing power obtained by each task, Num task (t) is the number of tasks in time slot t. gen (FLOPS) is the computing resource of the general-purpose server, and Num is the number of general-purpose servers in the data center.

[0105] Computing power scale constraints:

[0106] Computing power scale model

[0107] Meaning: The total computing power scale needs to consider the sum of the computing power scales corresponding to delay-tolerant and delay-sensitive tasks. The final computing power scale of the data center can be expressed as:

[0108] K total =K+K′

[0109] Where: K is the computing power scale corresponding to the delay-tolerant task. K′ is the computing power scale corresponding to the delay-sensitive task. Since delay-sensitive tasks cannot be scheduled, the maximum task computing power received by the data center is Assume that the average processing delay of each delay-sensitive task is t sen , we assume that t sen =0.1s. Then the computing power scale of a single data center is K′=Request sen / t sen .

[0110] C2: K′≤K max

[0111] Where: Constraints indicate that the task processing time must be before the tolerable time point and the data center size cannot exceed the maximum computing resource capacity. max Indicates the maximum computing capacity that a data center can support.

[0112] Computing power scale constraints:

[0113]

[0114] Regarding step 104:

[0115] In one embodiment of the present invention, the energy supply constraint is:

[0116] C2:E total ≤Q total

[0117] Where: Q total It is the total amount of energy that can be provided by the energy supplier.

[0118] Regarding step 106:

[0119] like Figure 5As shown, in one embodiment of the present invention, it can be seen from the above model that the decision variables include task allocation strategy, data center computing power scale strategy and data center computing power equipment construction strategy. Since the task allocation strategy is related to the computing power equipment configuration of the data center, and the computing power scale of the data center is the boundary condition of the computing power equipment construction strategy of the data center, the decisions are coupled with each other and it is difficult to optimize and solve them at the same time. It is necessary to split and transform the above variables. Considering the logical relationship of the variables, the task allocation strategy and the data center computing power scale strategy can be first determined based on the computing power requirements of the task. After obtaining the data center computing power scale strategy, the computing power equipment construction strategy is further obtained with the computing power scale as the boundary condition. It can be seen that the two-layer optimization model is suitable for solving the computing power planning model of this project.

[0120] The original optimization model was transformed into a multi-objective, two-layer joint optimization computing power allocation model. The upper-layer model serves as the main model, minimizing carbon emissions and maximizing profits to derive task allocation strategies and data center computing power scale strategies. The lower-layer model serves as a sub-model, using the data center computing power scale determined by the upper-layer model as the boundary condition and developing a data center computing power equipment construction plan with the goals of minimizing carbon emissions and maximizing profits.

[0121] The data center computing capacity scale obtained through optimization by the upper-level model is passed to the lower-level model. The lower-level model then optimizes the data center computing equipment deployment and passes the optimization targets of one iteration to the upper-level model for subsequent iterations. Through this top-to-bottom transfer, the two layers jointly solve the optimal computing capacity allocation strategy for the computing system.

[0122] 1) Upper model

[0123] The upper-level model is the main model for computing system planning. The decision variables are task allocation strategy and data center computing scale strategy. Its objective function is to minimize carbon emissions and electricity prices, as shown below.

[0124] The last generation normalizes the economic and environmental goals by adding weighted values ​​and summing them up. The comprehensive objective function is as follows.

[0125]

[0126]

[0127]

[0128] ω1+ω2=1

[0129] In the formula, ω1 and ω2 are weight coefficients, is the normalized value of the economic objective of the i-th solution in the last generation, is the normalized value of the environmental protection target of the i-th solution in the last generation, is the maximum value of the economic objectives among all solutions of the last generation, is the minimum value of the economic objective among all solutions of the last generation, is the maximum value of the environmental protection target among all solutions in the last generation, is the minimum value of the environmental protection target among all solutions of the last generation, and M is the number of solutions of the last generation.

[0130] After the upper-level model determines the scale of the data center, it is transmitted to the lower-level model to complete the configuration of the computing equipment in the data center.

[0131] (2) Lower-level model

[0132] The lower-level model is a submodel of the upper-level model, used to solve the subproblem of optimizing computing equipment deployment within the computing system. It coordinates the joint processing of tasks across different computing equipment based on their differences in cost, revenue, and environmental performance. The data center computing capacity scale of the upper-level model is passed to the lower-level model as its boundary condition.

[0133] The last generation normalizes the economic and environmental goals by adding weighted values ​​and summing them up. The comprehensive objective function is as follows.

[0134]

[0135]

[0136]

[0137] ω1+ω2=1

[0138] In the formula, ω1 and ω2 are weight coefficients, is the normalized value of the economic objective of the i-th solution in the last generation, is the normalized value of the environmental protection target of the i-th solution in the last generation, is the maximum value of the economic objectives among all solutions of the last generation, is the minimum value of the economic objective among all solutions of the last generation, is the maximum value of the environmental protection target among all solutions in the last generation, is the minimum value of the environmental protection target among all solutions of the last generation, and M is the number of solutions of the last generation.

[0139] During each optimization process, the upper-level model optimizes the computing power scale of the data center, and the lower-level model uses this as the boundary condition to optimize the optimal computing power configuration of the data center.

[0140] Regarding step 108:

[0141] In one embodiment of the present invention, the maximum profit during the planning period is determined by the following formula:

[0142]

[0143]

[0144]

[0145]

[0146] Cost e (t)=E(t)P(t)

[0147]

[0148]

[0149]

[0150]

[0151]

[0152] Income t =P task *T running

[0153]

[0154]

[0155]

[0156] TrueProfit Total =Profit Total *(1-λ)

[0157] In the formula, Cost Device Indicates the total cost of initial equipment purchase, Cost Device,l represents the total initial equipment acquisition cost of the data center, Indicates the number of type a CPU computing devices in the data center, Indicates the number of type b GPU computing devices in the data center, Indicates the number of c types of network devices in the data center, Indicates the number of d types of cabinets in the data center, Indicates the price of type a CPU computing equipment in the data center, Indicates the price of type b GPU computing equipment in the data center, represents the price of type c network equipment in the data center, Indicates the price of d types of cabinet equipment in the data center, represents the planned maintenance cost in year t, Indicates the number of type a CPU computing devices in the data center, Indicates the number of type b GPU computing devices in the data center, Indicates the number of c types of network devices in the data center, Indicates the number of d types of cabinets in the data center, represents the maintenance cost of type a CPU computing equipment in the data center in year t, represents the maintenance cost of type b GPU computing equipment in the data center in year t, represents the maintenance cost of type c network equipment in the data center in year t, represents the maintenance cost of type d cabinet equipment in the data center in year t, Indicates the annual equipment power consumption cost, HPR FJ A represents the heating price per unit area of ​​the data center. r,HFJ Indicates the heating area of ​​the data center, CPR FJ Table 2. Cooling price per unit area, A r,CFJ represents the cooling area of ​​the data center, w type represents the energy efficiency coefficient of the data center in different seasons, where w type The default value is 1.05, c e,t represents the terminal power supply price of the data center in year t, P request represents the energy consumption of the data center's computing power request equipment in year t, P basic represents the basic energy consumption of data center equipment in year t, Cost e,total is the total electricity price, T cycle is the life cycle in years, is the total electricity price of the data center corresponding to a typical day, T seai is the number of days of typical daily scei, Cost is the total electricity price of the data center within one day. e (t) The electricity price of data center m at time slot t,

[0158] represents the electricity price of the data center, PCI) represents the electricity price of the data center at time t, S t,vs is the value-added service revenue in year t, β vs The coefficient of the value-added service income ratio of the data center, The comprehensive energy purchase cost of the data center is Indicates the total cost of the data center, Energy t Indicates the total power consumption of all equipment in the data center in year t, Expense t Cost represents the construction cost of the data center in year t. Inv Indicates the initial investment cost, Cost Devicerepresents the total initial equipment acquisition cost of the data center, represents the total cost of the data center. Considering the delay in project planning and construction, the discount rate needs to be increased. r is the capital discount rate, and α is the data center loan ratio. is the financial cost in year t, is the financial cost of the data center in year t, rate is the annual loan interest rate of the data center loan, Cost Inv is the data center investment cost, α is the data center loan ratio, lim is the repayment period of the data center loan, Cost Total Planning the total cost of the integrated computing network system, Income t represents the computing power income in the tth year, P task represents the unit price of computing power in the data center in year t, T running represents the planned computing power demand at the center in year t, Residual is the total residual value of the computing network system equipment at the end of the planning period, H is the total number of equipment in the computing network system and H=A+B+C+D, β i is the residual value rate of the i-th equipment, Price i is the initial purchase price of the i-th device, Income Total is the total income of the network system over its entire life cycle, r is the capital discount rate, To calculate the economic benefits of the network system throughout its life cycle, TrueProfit Total is the total net profit over the entire life cycle, and λ is the income tax rate.

[0159] The upper limit constraint of equipment planning capacity is determined by the following formula:

[0160]

[0161]

[0162] Where, are the planned capacity upper limits of energy production equipment and energy storage equipment,

[0163] Equipment operation constraints include upper and lower output limits of energy supply equipment and charging and discharging power constraints of energy storage equipment.

[0164] Regarding step 110:

[0165] In one embodiment of the present invention, the power balance constraint includes an electric power balance constraint, a thermal power balance constraint, a cooling power balance constraint, a natural gas power balance constraint, and a hydrogen power balance constraint. The power balance constraint is determined by the following formula:

[0166] Electric power balance constraints

[0167]

[0168] Thermal power balance constraints

[0169]

[0170] Cold power balance constraint

[0171]

[0172] Natural gas power balance constraints

[0173]

[0174] Hydrogen power balance constraints

[0175]

[0176] Where Ω4 is the set of energy supply and production equipment that generates electricity, Ω5 is the set of energy storage equipment, Ω6 is the set of energy supply and production equipment that consumes electricity, and P e,m,t,s,h P is the electric power generated by the mth energy supply production equipment generating electricity in the typical day scenario h period in year t, e,j,ES-dis,t,s,h P is the discharge power of the jth energy storage device in the hth period of the typical day scenario in the tth year, e,k,t,s,h P is the power consumed by the kth energy supply device consuming electricity in the hth period of the typical day scenario in the tth year, eL,t,s,h is the other electric load in the typical day scenario h period in year t, P e,j,ES-ch,t,s,h P is the charging power of the jth energy storage device in the hth period of the typical day scenario in the tth year, cpeL,t,s,h is the power load consumed by computing equipment in the typical scenario of day s in year t during period h, Ω7 is the set of energy supply and production equipment that generates heat energy, Ω8 is the set of heat storage equipment, Ω9 is the set of energy supply and production equipment that consumes heat energy, and Q q,i,t,s,h Q is the thermal power generated by the i-th energy supply production equipment generating heat energy in the typical day scenario h period of year t, q,j,HS-dis,t,s,h Q is the heat release power of the jth heat storage device in the hth period of the typical day scenario in the tth year, q,k,t,s,h Q is the heat load of the kth energy supply production equipment consuming heat energy in the typical day scenario h period in year t, qL,t,s,h is the heat load of the typical day scenario h period in year t, Q q,j,HS-ch,t,s,h is the charging power of the jth heat storage device in the hth period of the typical day scenario in the tth year, Ω 10 A collection of energy supply production equipment for generating cold energy, Ω 11 For cold storage equipment collection, C c,i,t,s,h C is the cooling power generated by the i-th energy supply production equipment that generates cooling energy during the h-hour period in the typical day scenario of the t-th year, c,j,CS-dis,t,s,his the cooling power of the jth cold storage device in the hth period of the typical day scenario in the tth year, C cL,t,s,h is the cooling load of the typical day scenario h period in year t, C c,j,CS-ch,t,s,h is the cooling power of the jth cold storage device in the hth period of the typical day scenario in the tth year, Ω3 is the set of energy supply production equipment that consumes natural gas, and Ω 12 G is the gas storage equipment collection. g,t,s,h is the natural gas power purchased in the typical day scenario h period in year t, G g,i,GS-dis,t,s,h G is the gas discharge power of the jth gas storage device in the hth period of the typical day scenario in the tth year, g,j,t,s,h The natural gas power of the energy supply production equipment consuming natural gas at the jth natural gas consumption time period in the typical day scenario of the tth year, G g,i,GS-ch,t,s,h The charging power of the jth gas storage device in the hth period of the typical day scenario in the tth year, Ω 14 Energy supply production equipment collection for consuming hydrogen, Ω 13 For hydrogen storage equipment collection, is the amount of hydrogen purchased in the typical day scenario h period in year t, is the hydrogen discharge power of the jth hydrogen storage device in the hth period of the typical day scenario in the tth year, The hydrogen power of the energy supply production equipment consuming hydrogen energy at the jth time period in the typical day scenario of the tth year is: The hydrogen charging power of the jth hydrogen storage device in the hth period of the typical day scenario in the tth year.

[0177] The power exchange constraint between the energy system and the upper network is determined by the following formula:

[0178]

[0179]

[0180]

[0181]

[0182] 0≤P le,t,i,s,h ≤P legm,t,i,s,h

[0183] Where, are the minimum and maximum power purchases from the public grid thermal power, respectively; Buy gas power for minimum and maximum separately; Purchase hydrogen power for minimum and maximum respectively; are the minimum and maximum constant green electricity purchase power respectively; P le,t,i,s,h P is the purchase amount of green electricity in the i-th medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; legm,t,i,s,hThe amount of green electricity provided in the i-th green electricity medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year.

[0184] The equipment operation constraints are determined by the following formula:

[0185]

[0186]

[0187]

[0188]

[0189]

[0190]

[0191] Where, is the minimum value of the electricity provided by the i-th green electricity medium- and long-term transaction; The maximum electric power consumed by the kth energy supply device that consumes electric energy in year t; The maximum thermal power of the energy supply production equipment for the kth energy consuming heat in year t; The maximum natural gas power of the energy supply production equipment for the jth natural gas-consuming energy in year t; The maximum hydrogen power of the production equipment supplied by the energy source that consumes hydrogen in year t, j; are the rated capacities of the electricity, heat and cold energy production equipment in year t, They are the peak values ​​of other power loads, power load consumed by computing equipment, heat load, and cooling load in year t respectively.

[0192] Green electricity ratio constraints

[0193] GE rate ≥GE rate_nor

[0194] In the formula, GE rate is the proportion of green electricity, GE rate_nor The lower limit of the proportion of green electricity.

[0195] Regarding step 112:

[0196] like Figure 6As shown, in one embodiment of the present invention, it can be seen from the above model that the decision variables include equipment configuration strategy and system operation optimization variables. The system operation optimization problem takes the system equipment configuration strategy as the boundary condition. It is difficult to optimize the two at the same time. It is necessary to split and transform the two parts of the variables. Considering the logical relationship of the variables, the equipment configuration strategy can be determined first, and the system operation optimization decision can be made on this basis. The optimal solution is output through repeated iterations. It can be seen that the two-layer optimization model is suitable for solving the planning model of this project.

[0197] The original optimization model was converted into a multi-objective, two-layer joint optimization configuration model for an integrated energy system. The upper-layer model serves as the main model, representing the equipment selection and capacity optimization model. This model is used to select the optimal equipment configuration combination from the candidate equipment set, with the goal of minimizing total cost and pollutant emissions. The lower-layer model serves as the sub-model, representing the operation and scheduling optimization model. The equipment configuration scheme determined by the upper-layer model serves as the boundary condition, and is used to develop the optimal output plan for each device while maintaining regional energy supply and demand balance, with the goal of minimizing total operating cost.

[0198] The upper-level model optimizes the configuration strategies and planned capacities of various types of equipment and passes them to the lower-level model. The lower-level model optimizes the scheduling and operation of the integrated energy system and returns the optimal operation plan and total operating cost to the upper-level model. Through the optimization iteration of the upper and lower levels, the optimal configuration strategy of the energy system is obtained.

[0199] The upper-level model is the main model for energy system planning. The decision variables are the configuration and planned capacity of various types of equipment, namely the investment capacity of energy supply production equipment, the investment capacity of energy storage equipment, and the rated power of energy storage equipment. Its objective function is to minimize the total planning cost and the emission of pollutants.

[0200] By adding weighted values ​​and normalizing the economic and environmental goals and then summing them up, the comprehensive objective function is as follows.

[0201]

[0202]

[0203]

[0204] ω1+ω2=1

[0205] In the formula, ω1 and ω2 are weight coefficients, is the normalized value of the economic objective of the i-th solution in the current iteration, is the normalized value of the environmental protection target of the i-th solution in the current iteration, max i (f i,ntp ) is the maximum value of the economic objective among all solutions in the current iteration, min i (fi,ntp ) is the minimum value of the economic objective among all solutions in the current iteration, max i (f i,e ) is the maximum value of the environmental protection target among all solutions in the current iteration, min i (f i,e ) is the minimum value of the environmental protection target among all solutions in the current iteration, and M is the number of solutions in the current iteration.

[0206] After the lower-level model determines the operating strategy, it is transmitted to the upper-level model to complete the cost and benefit calculation related to the operating strategy.

[0207] (2) Lower-level model

[0208] The lower-level model is a submodel of the upper-level model, solving the subproblem of optimizing the scheduling of various energy supply and production equipment and energy storage devices within the energy system. Based on the varying supply and demand of different energy sources, it coordinates the operation of various energy supply and production equipment and the charging and discharging of energy storage devices to meet diverse energy demands. The configuration and planned capacity of various equipment generated by the upper-level model are transferred to the lower-level model as boundary conditions. Its objective function is to minimize total operating cost, as shown below.

[0209]

[0210] In the nth iteration of the optimization process, the upper-level model optimizes the configuration and planned capacity of various equipment and passes it to the lower-level model. The lower-level model uses this as the boundary condition to optimize the optimal scheduling operation plan, and returns to the upper level to calculate the profit and pollutant gas emissions during the planning period as the initial values ​​for the n+1th iteration.

[0211] Regarding step 114:

[0212] like Figure 4 As shown, in one embodiment of the present invention, the data center's energy supply is jointly provided by a local integrated energy station and external power grids, natural gas grids, and hydrogen networks. The multi-objective computing power planning and design aims to minimize carbon emissions, maximize profits, and minimize energy consumption. Using the data center's various energy supply curves and prices as boundary conditions, the computing power tasks are time-optimized to determine the data center's initial scale and computing power load demand curve. The energy multi-objective planning and design module calculates the data center's cooling load, heating load, and other electrical loads based on the computing power load demand curve. Using these as boundary conditions, and with the goals of maximizing profits and minimizing carbon emissions during the planning period, the module optimizes the data center's energy supply system's energy supply facility capacity, the scale of public grid thermal power and green power transactions, and the scale of natural gas and hydrogen transactions. The multi-objective computing power and energy planning and design utilize computing power tasks and energy supply as coupling points and boundary interactions. Through multiple iterations, the optimal planning scheme for computing, network, and energy integration is ultimately optimized.

[0213] like Figure 2 、 Figure 3 As shown, the embodiment of the present invention provides a computing network collaborative planning device that considers the timing scheduling of computing tasks. The device embodiment can be implemented by software, or by hardware or a combination of software and hardware. From the hardware level, Figure 2 As shown in the figure, a hardware architecture diagram of an electronic device where a computing network collaborative planning device is located that considers the timing scheduling of computing tasks provided by an embodiment of the present invention is provided. Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, the electronic device in the embodiment may also include other hardware, such as a forwarding chip responsible for processing messages, etc. Taking software implementation as an example, Figure 3 As shown, as a device in a logical sense, it is formed by the CPU of the electronic device in which it is located reading the corresponding computer program in the non-volatile memory into the internal memory and running it.

[0214] like Figure 3 As shown, this embodiment provides a computing network collaborative planning device that considers computing task timing scheduling, and the device includes:

[0215] A model building module 300 is used to establish a computing power network planning model and an energy supply planning model; wherein the computing power network planning model includes a first upper layer model and a first lower layer model, and the energy supply planning model includes a second upper layer model and a second lower layer model;

[0216] A first data processing module 302 is configured to determine the center computing power scale based on the objective function of maximizing data center revenue and minimizing carbon emissions, and based on the constraints of latency and computing power scale, and to send the center computing power scale to the first lower-layer model.

[0217] A second data processing module 304 is configured to determine the carbon emissions and profits of the data center using the first lower-layer model as objective functions, with energy constraints and computing power scale constraints as constraints, and to send the carbon emissions and profits of the data center to the first upper-layer model;

[0218] A third data processing module 306 is configured to repeatedly iterate the first upper-layer model and the first lower-layer model to determine a first optimal solution; wherein the first optimal solution includes a computing power equipment planning scheme and a computing power energy consumption load curve, and the computing power energy consumption load curve is used as an input of an energy supply planning model for coupled interaction;

[0219] A fourth data processing module 308 is configured for the second upper-layer model to determine the equipment configuration strategy and planned capacity with the maximum profit and the minimum carbon emission during the planning period as the objective function, and the equipment planned capacity upper limit constraint and the equipment commissioning constraint as the constraint conditions, and to send the equipment configuration strategy and planned capacity to the second lower-layer model;

[0220] A fifth data processing module 310 is configured to determine the optimal operation plan and total operation cost for the second lower-layer model using minimization of operation cost as an objective function, and using power balance constraints, equipment operation constraints, power exchange constraints between the energy system and the upper-level network, and green power ratio constraints as constraints, and to send the optimal operation plan and total operation cost to the second upper-layer model;

[0221] a sixth data processing module 312 configured to repeatedly iterate the second upper-layer model and the second lower-layer model to determine a second optimal solution; wherein the second optimal solution includes an energy supply planning scheme, a power curve of power generation equipment, a public thermal power curve, and a green power trading curve, and the power curve of power generation equipment, the public thermal power curve, and the green power trading curve are used as inputs of a computing power network planning model for coupled interaction;

[0222] The loop coupling interaction module 314 is used for loop coupling interaction between the computing power network planning model and the energy supply planning model, and has realized the computing power network collaborative planning of the computing power task timing scheduling.

[0223] In an embodiment of the present invention, the model building module 300 can be used to execute step 100 in the above method embodiment, the first data processing module 302 can be used to execute step 102 in the above method embodiment, the second data processing module 304 can be used to execute step 104 in the above method embodiment, the third data processing module 306 can be used to execute step 106 in the above method embodiment, the fourth data processing module 308 can be used to execute step 108 in the above method embodiment, the fifth data processing module 310 can be used to execute step 110 in the above method embodiment, the sixth data processing module 312 can be used to execute step 112 in the above method embodiment, and the loop coupling interaction module 314 can be used to execute step 114 in the above method embodiment.

Claims

1. A computing network collaborative planning method considering the timing scheduling of computing tasks, characterized by: The method is applied to a single-node computing network and includes: Establishing a computing power network planning model and an energy supply planning model; wherein the computing power network planning model includes a first upper layer model and a first lower layer model, and the energy supply planning model includes a second upper layer model and a second lower layer model; The first upper-layer model takes maximizing data center revenue and minimizing carbon emissions as objective functions, and takes latency constraints and computing power scale constraints as constraints, determines the center computing power scale, and sends the center computing power scale to the first lower-layer model; The first lower-layer model takes maximizing data center revenue and minimizing carbon emissions as objective functions, and takes energy constraints and computing power scale constraints as constraints, determines the data center's carbon emissions and profits, and sends the data center's carbon emissions and profits to the first upper-layer model; The first upper-layer model and the first lower-layer model are repeatedly iterated to determine a first optimal solution; wherein the first optimal solution includes a computing power equipment planning scheme and a computing power energy consumption load curve, and the computing power energy consumption load curve is used as an input of the energy supply planning model to perform coupled interaction; The second upper-layer model takes maximizing profit and minimizing carbon emissions during the planning period as objective functions, and takes equipment planning capacity upper limit constraints and equipment commissioning constraints as constraints, determines equipment configuration strategies and planned capacity, and sends the equipment configuration strategies and planned capacity to the second lower-layer model; The second lower-layer model uses minimizing the operating cost as the objective function, and uses power balance constraints, equipment operation constraints, power exchange constraints between the energy system and the upper network, and green power ratio constraints as constraints to determine the optimal operating plan and total operating cost, and sends the optimal operating plan and total operating cost to the second upper-layer model; The second upper-layer model and the second lower-layer model are repeatedly iterated to determine a second optimal solution; wherein the second optimal solution includes an energy supply planning scheme, a power curve of power generation equipment, a public grid thermal power curve, and a green electricity trading curve, and the power curve of power generation equipment, the public grid thermal power curve, and the green electricity trading curve are used as inputs of the computing power network planning model for coupled interaction; The computing power network planning model and the energy supply planning model are cyclically coupled and interacted to realize the computing network collaborative planning of computing power task timing scheduling.

2. The method according to claim 1, characterized in that The objective function is to maximize the data center's revenue and minimize its carbon emissions, including the following formula: max f ntp S t,c =C t,c ·(E t,nom -E t,co2 ) / 10000 f ep =f in -f c f tp =f ep -f ep ·l vat ·l umt -f ep ·l vat ·l est f ntp =(f ep -f ep ·l vat ·l umt -f ep ·l vat ·l est )·(1-λ int ) Where, The number of energy supply production equipment put into operation in year t; The capacity of the selected model of production equipment for the i-th energy supply; is the number of the i-th energy storage device put into operation in year t; is the capacity of the model selected for the i-th energy storage device; C inv is the initial investment cost; T is the life cycle; r is the capital discount rate; is the investment cost in year t; α lr,t is the loan ratio in year t; i is the number of energy supply production equipment; Ω1 is the set of energy supply production equipment; The unit capacity investment cost of the production equipment for the i-th energy supply; is the investment and construction capacity of the i-th energy supply production equipment in year t; j is the energy storage equipment number; Ω2 is the energy storage equipment set; is the unit capacity investment cost of the j-th energy storage device; is the investment and construction capacity of the j-th energy storage equipment in year t, The annual operation and maintenance cost per unit capacity of the production equipment supplying the i-th energy source; is the annual operation and maintenance cost per unit capacity of the j-th energy storage device, is the purchase cost of thermal power from the public grid in year t, is the medium- and long-term transaction cost of the i-th green electricity in year t, is the normal green electricity purchase cost in year t, is the penalty cost for the deviation of the i-th green power medium- and long-term transaction in the t-th year, is the natural gas purchase cost in year t, is the cost of purchasing hydrogen in year t, C e,t,s,h P is the unit price of electricity purchased from the public grid during the h period of the typical day scenario in year t; e,t,s,h is the amount of electricity purchased from the public grid during the h period of the typical day scenario in year t, C le,t,i,s,h P is the purchase price of green electricity in the i-th medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; le,t,i,s,h C is the purchase amount of green electricity in the i-th medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year, ge,t,s,h P is the normal green electricity purchase price for the typical day scenario h period in year t; ge,t,s,h is the normal green electricity purchase amount in the h period of the typical day scenario in year t, C g,t,s,h G is the natural gas purchase price in the typical day scenario h period in year t; g,t,s,h is the natural gas purchase amount in the typical day scenario h period in year t, The unit price of hydrogen purchased during the h period of the typical day scenario in year t is: is the amount of hydrogen purchased in the typical day scenario h period in year t, C le,t,i,s,h,dp The penalty price for deviation of green electricity medium- and long-term transaction in the i-th green electricity medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; P le,t,i,s,h,dp P is the deviation of the i-th green electricity medium- and long-term transaction caused by the responsibility of the power purchaser in the h-th period of the typical day scenario in the t-th year; lest,t,i,s,h The transaction volume of the i-th green power medium- and long-term trading contract in the h-th period of the typical day scenario in the t-th year; P legm,t,i,s,h The amount of green electricity provided by medium- and long-term transactions in the i-th period of time in the typical day scenario of the t-th year is C fc,t is the financial cost in year t; r i,tt is the annual loan interest rate for the loan in year tt; C inv,tt is the investment cost in year tt; α lr,tt is the loan ratio in year tt; N tt is the repayment period of the loan in year tt, f c is the total investment cost of the integrated energy system; f t,c is the investment cost in year t, SR t is the energy supply income in year t, C e,t,s,h,S are the electricity selling price in the typical day scenario h period in year t; P eL,t,s,h P is the other electric load in the typical day scenario h period in year t; cpeL,t,s,h HPR is the power load consumed by computing equipment in the typical day scenario h period in year t; FJ is the heating price per unit area; CPR FJ is the cooling price per unit area; A r,HFJ is the heating area; A r,CFJ is the cooling area, S t,up is the surplus access income in year t; S is the number of typical daily scenarios each year; N s is the duration of the typical day scene s in a year; H is the total number of time periods in a day, C up,e,t,s,h P is the unit price of surplus electricity on the grid in the h period of the typical day scenario in year t; up,e,t,s,h is the on-grid power consumption in the h period of the typical day scenario in year t, where: S inv Subsidize total investment; is the investment subsidy for year t; Subsidy for initial investment per unit capacity of the i-th energy supply production equipment; is the initial investment subsidy per unit capacity of the j-th energy storage device; S t,sub is the power generation subsidy income in year t; S is the number of typical day scenarios each year; N s is the duration of the typical day scene s in a year; H is the total number of time periods in a day, C wt,e,t,s,h 、C pv,q,t,s,h are the electricity subsidies for wind power generation and photovoltaic power generation in the h period of the typical day scenario in year t; P wt,e,t,s,h 、P pv,e,t,s,h are the wind power generation and photovoltaic power generation in the typical day scenario h period in year t, S t,c is the environmental benefit in year t; C t,c is the carbon trading price in year t; E t,nom is the carbon quota for year t; E t,co2 is the carbon dioxide emissions in year t; α e,CO2 S is the emission coefficient of carbon dioxide generated by the consumption of electricity; t,i,le,dp is the deviation income of the i-th green power medium- and long-term transaction in the t-th year, E RV is the total residual value of the integrated energy system equipment at the end of the planning period, M n is the total number of equipment in the integrated energy system, δ i is the residual value rate of the i-th equipment, C INV,i is the initial investment of the i-th equipment, f in is the total benefit of the integrated energy system over its entire life cycle; f t,in is the total income in year t, f ep For comprehensive energy services full life cycle economic benefits, f ntp is the total net profit over the entire life cycle, λ vat is the value-added tax rate, λ umt Maintaining tax rates for urban construction, λ est is the education surcharge rate, which is 5%, λ int is the income tax rate, f ntp is the total net profit over the entire life cycle, f e is carbon dioxide emissions, is the CO2 emission in year t; N s is the duration of the typical day scene s in a year; is the carbon dioxide emission coefficient produced by natural gas.

3. The method according to claim 2, characterized in that The maximum profit during the planning period is determined by the following formula: Cost e (t)=E(t)P(t) Income t =P task *T running TrueProfit Total =Profit Total *(1-λ) In the formula, Cost Device Indicates the total cost of initial equipment purchase, Cost Device,l represents the total initial equipment acquisition cost of the data center, Indicates the number of type a CPU computing devices in the data center, Indicates the number of type b GPU computing devices in the data center, Indicates the number of c types of network devices in the data center, Indicates the number of d types of cabinets in the data center, Indicates the price of type a CPU computing equipment in the data center, Indicates the price of type b GPU computing equipment in the data center, represents the price of type c network equipment in the data center, Indicates the price of d types of cabinet equipment in the data center, represents the planned maintenance cost in year t, represents the maintenance cost of type a CPU computing equipment in the data center in year t, represents the maintenance cost of type b GPU computing equipment in the data center in year t, represents the maintenance cost of type c network equipment in the data center in year t, represents the maintenance cost of type d cabinet equipment in the data center in year t, Indicates the annual equipment power consumption cost, HPR FJ A represents the heating price per unit area of ​​the data center. r,HFJ Indicates the heating area of ​​the data center, CPR FJ Table 2. Cooling price per unit area, A r,CFJ represents the cooling area of ​​the data center, w type represents the energy efficiency coefficient of the data center in different seasons, where w type The default value is 1.05, c e,t represents the terminal power supply price of the data center in year t, P request represents the energy consumption of the data center's computing power request equipment in year t, P basic represents the basic energy consumption of data center equipment in year t, Cost e，total is the total electricity price, T cycle is the life cycle in years, is the total electricity price of the data center corresponding to the typical day scei, T scei is the number of days of typical daily scei, Cost e (t) The electricity price of data center m at time slot t, represents the electricity price of the data center, P(t) represents the electricity price of the data center at time t, S t,vs is the value-added service revenue in year t, β vs The coefficient of the value-added service income ratio of the data center, The comprehensive energy purchase cost of the data center is Indicates the total cost of the data center, Energy t Indicates the total power consumption of all equipment in the data center in year t, Expense t Cost represents the construction cost of the data center in year t. Inv Indicates the initial investment cost, Cost Device represents the total initial equipment acquisition cost of the data center, The total cost of the data center is shown in Figure 2. Considering the delays in project planning and construction, the discount rate needs to be increased. is the financial cost in year t, is the financial cost of the data center in year t, rate is the annual loan interest rate of the data center loan, Cost Inv is the data center investment cost, α is the data center loan ratio, lim is the repayment period of the data center loan, Cost Total Planning the total cost of the integrated computing network system, Income t represents the computing power income in the tth year, P task represents the unit price of computing power in the data center in year t, T running represents the planned computing power demand at the center in year t, Residual is the total residual value of the computing network system equipment at the end of the planning period, H is the total number of equipment in the computing network system and H=A+B+C+D, β i is the residual value rate of the i-th equipment, Price i is the initial purchase price of the i-th device, Income Total is the total income of the network system over its entire life cycle, r is the capital discount rate, To calculate the economic benefits of the network system throughout its life cycle, TrueProfit Total is the total net profit over the entire life cycle, and λ is the income tax rate.

4. The method according to claim 3, characterized in that The power balance constraints include electric power balance constraints, thermal power balance constraints, cooling power balance constraints, natural gas power balance constraints, and hydrogen power balance constraints. The power balance constraints are determined by the following formula: Where Ω4 is the set of energy supply and production equipment that generates electricity, Ω5 is the set of energy storage equipment, Ω6 is the set of energy supply and production equipment that consumes electricity, and P e,m,t,s,h P is the electric power generated by the mth energy supply production equipment generating electricity in the typical day scenario h period in year t, e,j,ES-dis,t,s,h P is the discharge power of the jth energy storage device in the hth period of the typical day scenario in the tth year, e,k,t,s,h P is the power consumed by the kth energy supply device consuming electricity in the hth period of the typical day scenario in the tth year, eL,t,s,h is the other electric load in the typical day scenario h period in year t, P e,j,ES-ch,t,s,h P is the charging power of the jth energy storage device in the hth period of the typical day scenario in the tth year, cpeL,t,s,h is the power load consumed by computing equipment in the typical scenario of day s in year t during period h, Ω7 is the set of energy supply and production equipment that generates heat energy, Ω8 is the set of heat storage equipment, Ω9 is the set of energy supply and production equipment that consumes heat energy, and Q q,i,t,s,h Q is the thermal power generated by the i-th energy supply production equipment generating heat energy in the typical day scenario h period of year t, q,j,HS-dis,t,s,h Q is the heat release power of the jth heat storage device in the hth period of the typical day scenario in the tth year, q,k,t,s,h Q is the heat load of the kth energy supply production equipment consuming heat energy in the typical day scenario h period in year t, qL,t,s,h is the heat load of the typical day scenario h period in year t, Q q,j,HS-ch,t,s,h is the charging power of the jth heat storage device in the hth period of the typical day scenario in the tth year, Ω 10 A collection of energy supply production equipment for generating cold energy, Ω 11 For cold storage equipment collection, C c,i,t,s,h C is the cooling power generated by the i-th energy supply production equipment that generates cooling energy during the h-hour period in the typical day scenario of the t-th year, c,j,CS-dis,t,s,h is the cooling power of the jth cold storage device in the hth period of the typical day scenario in the tth year, C cL,t,s,h is the cooling load of the typical day scenario h period in year t, C c,j,CS-ch,t,s,h is the cooling power of the jth cold storage device in the hth period of the typical day scenario in the tth year, Ω3 is the set of energy supply production equipment that consumes natural gas, and Ω 12 G is the gas storage equipment collection. g,t,s,h is the natural gas power purchased in the typical day scenario h period in year t, G g,i,GS-dis,t,s,h G is the gas discharge power of the jth gas storage device in the hth period of the typical day scenario in the tth year, g,j,t,s,h The natural gas power of the energy supply production equipment consuming natural gas at the jth natural gas consumption time period in the typical day scenario of the tth year, G g,i,GS-ch,t,s,h The charging power of the jth gas storage device in the hth period of the typical day scenario in the tth year, Ω 14 Energy supply production equipment collection for consuming hydrogen, Ω 13 For hydrogen storage equipment collection, is the amount of hydrogen purchased in the typical day scenario h period in year t, is the hydrogen discharge power of the jth hydrogen storage device in the hth period of the typical day scenario in the tth year, The hydrogen power of the energy supply production equipment consuming hydrogen energy at the jth time period in the typical day scenario of the tth year is: The hydrogen charging power of the jth hydrogen storage device in the hth period of the typical day scenario in the tth year.

5. The method according to claim 4, characterized in that The upper limit constraint of the equipment planning capacity is determined by the following formula: Where, They are the planned capacity upper limits of energy production equipment and energy storage equipment respectively.

6. The method according to claim 5, characterized in that The power exchange constraint between the energy system and the upper network is determined by the following formula: 0≤P le,t,i,s,h ≤P legm,t,i,s,h Where, are the minimum and maximum power purchases from the public grid thermal power, respectively; Buy gas power for minimum and maximum separately; Purchase hydrogen power for minimum and maximum respectively; are the minimum and maximum constant green electricity purchase power respectively; P le,t,i,s,h P is the purchase amount of green electricity in the i-th medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year; legm,t,i,s,h The amount of green electricity provided in the i-th green electricity medium- and long-term transaction in the h-th period of the typical day scenario in the t-th year.

7. The method according to claim 6, characterized in that The equipment operation constraints are determined by the following formula: Where, is the minimum value of the electricity provided by the i-th green electricity medium- and long-term transaction; The maximum electric power consumed by the kth energy supply device that consumes electric energy in year t; The maximum thermal power of the energy supply production equipment for the kth energy consuming heat in year t; The maximum natural gas power of the energy supply production equipment for the jth natural gas-consuming energy in year t; The maximum hydrogen power of the production equipment supplied by the energy source that consumes hydrogen in year t, j; are the rated capacities of the electricity, heat and cold energy production equipment in year t, They are the peak values ​​of other power loads, power load consumed by computing equipment, heat load, and cooling load in year t respectively.

8. A computing network collaborative planning device that considers the timing scheduling of computing tasks, characterized by: include: A model building module, configured to establish a computing power network planning model and an energy supply planning model; wherein the computing power network planning model includes a first upper layer model and a first lower layer model, and the energy supply planning model includes a second upper layer model and a second lower layer model; A first data processing module is configured to determine the center computing power scale based on the objective function of maximizing data center revenue and minimizing carbon emissions, and based on the constraints of latency and computing power scale, and to send the center computing power scale to the first lower-layer model; a second data processing module configured to determine the carbon emissions and profits of the data center using the first lower-layer model as objective functions and energy constraints and computing power scale constraints as constraints, and to send the carbon emissions and profits of the data center to the first upper-layer model; a third data processing module configured to repeatedly iterate the first upper-layer model and the first lower-layer model to determine a first optimal solution; wherein the first optimal solution includes a computing power equipment planning scheme and a computing power energy consumption load curve, and the computing power energy consumption load curve is used as an input of an energy supply planning model for coupled interaction; A fourth data processing module is configured for the second upper-layer model to determine the equipment configuration strategy and planned capacity with the maximum profit and the minimum carbon emission during the planning period as the objective function, and with the equipment planned capacity upper limit constraint and the equipment commissioning constraint as the constraint conditions, and to send the equipment configuration strategy and planned capacity to the second lower-layer model; A fifth data processing module is configured to determine the optimal operation plan and total operation cost for the second lower-layer model using minimization of operation cost as the objective function, and using power balance constraints, equipment operation constraints, power exchange constraints between the energy system and the upper-level network, and green power ratio constraints as constraints, and to send the optimal operation plan and total operation cost to the second upper-layer model; a sixth data processing module configured to iteratively perform the second upper-layer model and the second lower-layer model to determine a second optimal solution; wherein the second optimal solution includes an energy supply planning scheme, a power curve of power generation equipment, a public thermal power curve, and a green power trading curve, and the power curve of power generation equipment, the public thermal power curve, and the green power trading curve are used as inputs of a computing power network planning model for coupled interaction; The cyclic coupling interaction module is used for cyclic coupling interaction between the computing power network planning model and the energy supply planning model, and has realized the computing power network collaborative planning of the computing power task timing scheduling.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

Images