Method for energy efficiency optimization of data centers considering multi-period device coupling

Abstract

This invention discloses a data center energy efficiency optimization method considering multi-time period device coupling, comprising the following steps: 1) establishing a data center equipment energy consumption model; 2) establishing a data center energy efficiency optimization model based on multi-time period device coupling; 3) solving the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period device coupling to obtain the computing power allocation and start / stop scheme of data center IT equipment, as well as the air supply temperature of air conditioning. This invention establishes a data center energy efficiency optimization model based on multi-time period device coupling. Based on the respective energy consumption models of IT and air conditioning equipment, it comprehensively considers the coupling relationship between data center IT and air conditioning equipment as well as the coupling relationship between different time periods, and establishes a data center energy efficiency optimization model based on multi-time period device coupling with the total cost of the data center as the objective function.

Description

Technical Field

[0001] This invention relates to the field of data center energy efficiency optimization, specifically a data center energy efficiency optimization method that considers multi-time period equipment coupling. Background Technology

[0002] A data center refers to specific network devices used to transmit, accelerate, display, compute, and store data information on the Internet infrastructure. Currently, technologies such as the Internet of Things (IoT), cloud computing, and artificial intelligence are widely used throughout society, and data centers have become the cornerstone of global economic development in today's big data era. However, this has also brought about serious high energy consumption problems. Taking my country as an example, in 2020, the total electricity consumption of data centers in my country was 204.5 billion kilowatt-hours, accounting for approximately 2.7% of the total social electricity consumption, while the annual power generation of the Three Gorges Dam in 2020 was only 111.8 billion kilowatt-hours. It is estimated that by 2025, the total electricity consumption of data centers nationwide will reach 395.2 billion kilowatt-hours, accounting for approximately 4.1% of the total social electricity consumption. Power Usage Effectiveness (PUE) is a core indicator for measuring the energy efficiency of data centers. It is defined as the ratio of the total energy consumption of data center equipment to the total energy consumption of IT equipment. The lower the PUE value, the higher the energy efficiency. Currently, the PUE of domestic data centers is generally 1.5 to 2.0, which is significantly lower than the international advanced level of 1.0 to 1.5, indicating substantial room for energy conservation. The enormous energy consumption of data centers severely hinders the creation of an energy-efficient society and the achievement of "dual carbon" goals, necessitating urgent efforts to improve the energy efficiency of data centers and achieve green and sustainable development.

[0003] Currently, the common approach to data center energy efficiency optimization is to analyze the operating characteristics of IT and air conditioning equipment, establish a single-period energy consumption model for the data center, and then transform the data center energy efficiency optimization problem into an optimization problem with several constraints. Various optimization algorithms are then used to find the optimal PUE (Power Usage Effectiveness) scheme for IT equipment computing power allocation and start / stop, as well as the air supply temperature of air conditioning equipment, under the conditions of meeting total computing power requirements and normal equipment operation. This ultimately optimizes the allocation of computing power and start / stop of IT equipment, and the air supply temperature of air conditioning equipment, reducing the total energy consumption of the data center. However, existing models only consider single-period energy efficiency optimization, ignoring the inter-period coupling relationship between IT and air conditioning equipment. Essentially, the model is treated as multiple independent single-period optimizations. The optimal strategy provided by this type of optimization method may lead to problems such as excessive equipment wear or unmet computing power requirements in actual implementation. Therefore, when modeling data center energy efficiency optimization, it is essential to consider the inter-period coupling of energy efficiency. Through multi-period efficient operation management of the data center, the operating status of IT equipment should be dynamically adjusted, and inefficient IT equipment should be shut down or put into hibernation to improve overall energy efficiency and achieve green and sustainable development of the data center. In summary, it is necessary to comprehensively consider the energy consumption characteristics and heat exchange coupling relationship of data center IT and air conditioning equipment, as well as the start-up, shutdown, and ramp-up constraints of IT equipment during different time periods, and propose a multi-period energy efficiency optimization method for data centers based on equipment coupling. Summary of the Invention

[0004] The purpose of this invention is to provide a data center energy efficiency optimization method that considers multi-time period device coupling, comprising the following steps:

[0005] 1) Establish an energy consumption model for data center equipment;

[0006] The data center equipment energy consumption model includes an IT equipment energy consumption model, an air conditioning equipment energy consumption model, and a convection heat transfer model.

[0007] The energy consumption models for IT equipment are shown below:

[0008] P 计算 =(P max -P idle )u+P idle (1)

[0009] f(T IT )=b1+b2T IT (2)

[0010] P IT =[(P max -P idle )u+P idle ](b1+b2T IT (3)

[0011] In the formula, max and idle represent the full-load and idle states of the IT equipment, respectively; (P max -P idle u represents dynamic power consumption; P idle P represents static power; max Represents full-load power; u∈[0,1] represents the resource utilization rate of IT equipment; P 计算 f(T) represents the computing power consumption of IT equipment. IT () represents the leakage power consumption of the IT equipment; b1 and b2 are constants; T IT It refers to the chip temperature of IT equipment; P IT This indicates the energy consumption of IT equipment.

[0012] The energy consumption model for air conditioning equipment is shown below:

[0013]

[0014] In the formula, k T S represents the ambient temperature coefficient. dc P represents the sum of the projected areas of IT equipment; aircon Energy consumption of air conditioning equipment; P IT Indicates the energy consumption of IT equipment;

[0015] Among them, the total air conditioning cooling load Q aircon As shown below:

[0016] Q aircon =Q1+Q2 (5)

[0017] The heat generated by IT equipment (Q1) and the heat generated by air conditioning equipment (Q2) are shown below:

[0018] Q1 = 97% P IT (6)

[0019] Q2 = k T S dc (7)

[0020] The coefficient of performance (COP) of the air conditioner is shown below:

[0021]

[0022] In the formula, T out The air supply temperature for air conditioning equipment.

[0023] The convective heat transfer model is shown below:

[0024] T IT =P IT R in +T in (9)

[0025] T in =T out +97% DP IT (10)

[0026] In the formula, D is the heat transfer coefficient; T out The supply air temperature for air conditioning equipment; T in R is the inlet airflow temperature of the IT equipment. in The equivalent thermal resistance for inlet convection heat transfer.

[0027] 2) Establish a data center energy efficiency optimization model based on multi-time period device coupling;

[0028] The objective function of the data center energy efficiency optimization model based on multi-time period device coupling is as follows:

[0029]

[0030] In the formula, FP total (t)Δt represents the operating cost of the data center; P total (t) represents the total power of the data center during time period t; F is the operating cost coefficient; I i (t)[1-I i (t-1)]S i This refers to the start-up and shutdown costs of IT equipment; t represents the current time period; T is the total number of time periods; Δt is the time interval; I i (t) represents the operating state of the i-th IT device during time period t, where 0 indicates it is stopped and 1 indicates it is powered on or in standby mode; S i It is the start-up and shutdown cost coefficient.

[0031] The constraints of the data center energy efficiency optimization model based on multi-time period equipment coupling include computing power demand constraints, air conditioning supply air temperature range constraints, chip temperature upper limit constraints, inlet airflow temperature range constraints, IT equipment ramp-up constraints, and IT equipment downtime constraints.

[0032] The computing power requirement constraints are as follows:

[0033]

[0034] In the formula, u i It is the total computing power of the i-th IT device; u total This refers to the computing power currently provided by the data center; D IT This represents the computing power requirement; N is the number of IT devices.

[0035] The air conditioning supply air temperature range constraints are as follows:

[0036] T out,min ≤T out (t)≤T out,max(13)

[0037] In the formula, T out,max and T out,min These are the upper and lower limits of the air conditioner's air supply temperature; T out (t) is the air conditioner supply temperature at time t;

[0038] The upper limit constraint of chip temperature is as follows:

[0039] T IT (t)≤T IT,max (14)

[0040] In the formula, T IT,max This refers to the upper temperature limit for the normal operation of chips in IT equipment.

[0041] The inlet airflow temperature range constraints are as follows:

[0042] 18℃≤T in ≤27℃ (15)

[0043] The ramp-up constraints for IT equipment are as follows:

[0044]

[0045] In the formula, R represents the power consumption of the i-th IT device during time period t when the device is powered on; IT This represents the gradient coefficient.

[0046] 3) Solve the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period equipment coupling to obtain the computing power allocation and start-up / shutdown scheme of data center IT equipment, as well as the air supply temperature of air conditioners.

[0047] The steps for solving the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period equipment coupling include:

[0048] 3.1) By simplifying the IT equipment accordingly, we get:

[0049]

[0050]

[0051] R IT,eq,i =n i ·R IT,i (19)

[0052] In the formula, u eq,i P IT,eq,i and R IT,eq,i These are the equivalent IT equipment computing power, equivalent IT equipment power consumption, and equivalent IT equipment ramp rate; uj Let n be the total computing power of the j-th IT device; i P represents the total number of IT devices. IT,j R represents the power consumption of the j-th IT device. IT,i For IT equipment ramp-up coefficient;

[0053] 3.2) Equivalent IT equipment computing power u eq,i Equivalent IT equipment power consumption P IT,eq,i And equivalent IT equipment ramp factor R IT,eq,i Substituting these into the data center equipment energy consumption model, we obtain the equivalent IT equipment energy consumption model, the equivalent air conditioning equipment energy consumption model, and the equivalent convection heat transfer model.

[0054] 3.3) Based on step 3.2), update the data center energy efficiency optimization model based on multi-time period device coupling;

[0055] 3.4) The look-ahead decoupling algorithm is used to decompose the inter-time coupling constraints, resulting in a simplified data center energy efficiency optimization model; the decomposed inter-time coupling constraints are shown below:

[0056]

[0057] N f ={i|N i ≤m}, 0<m<t (21)

[0058] N s ={i|N i >m}, 0 < m < t (22)

[0059]

[0060]

[0061] In the formula, N i Let m be the minimum number of time intervals required for the power of IT device i to increase from the lower limit to the upper limit; m is any integer; N f It is a rapidly ascending set; N s It is a slowly ascending set; It is the maximum power consumption that all IT devices can increase during the m time periods from time period t-m to time period t; This is the maximum capacity that the computing power of all IT devices can increase during this period. P represents the total computing power requirement of the data center during time period t. IT,imax P IT,imin Δt represents the maximum and minimum power consumption of the IT equipment; Δt represents the time difference.

[0062] The decoupling conditions for the inter-time coupling constraints after decomposition include: for any m, all satisfy...

[0063] 4) Use the branch and bound method to solve the simplified data center energy efficiency optimization model, and obtain the computing power allocation and start-up / shutdown scheme of IT equipment, as well as the final optimization results of the air supply temperature of the air conditioner.

[0064] It is worth noting that the technical solution of this invention first establishes the energy consumption characteristics and heat exchange coupling relationship of data center IT and air conditioning equipment, and establishes individual energy consumption and heat exchange models for IT and air conditioning equipment. Then, it introduces inter-time period coupling relationships to establish a data center energy efficiency optimization model based on multi-time period equipment coupling, achieving multi-time period energy efficiency optimization of the data center. Based on look-ahead decoupling and equipment aggregation algorithms, a multi-time period data center energy efficiency optimization solution method is proposed, providing theoretical support for data center energy efficiency management system-level modeling and operational optimization.

[0065] The technical effects of this invention are undeniable, and its beneficial effects are as follows:

[0066] This invention establishes a data center energy efficiency optimization model based on multi-time period equipment coupling. Based on the energy consumption models of IT and air conditioning equipment, it comprehensively considers the coupling relationship between IT and air conditioning equipment in the data center as well as the coupling relationship between time periods, and establishes a data center energy efficiency optimization model based on multi-time period equipment coupling with the total cost of the data center as the objective function.

[0067] This invention proposes a data center energy efficiency optimization algorithm based on multi-time period device coupling. Addressing the challenges posed by time period coupling and IT equipment combination, this invention proposes a data center energy efficiency optimization algorithm based on look-ahead decoupling algorithm and IT equipment aggregation algorithm. Integer variables are processed through equipment equivalence, and the look-ahead decoupling algorithm reduces the number of times inter-time period coupling constraints take effect. Finally, the algorithm outputs the final optimized results of IT equipment computing power allocation and start / stop scheme, as well as air conditioning supply temperature.

[0068] This invention has strong potential for industrial applications. Existing research often neglects the time-period coupling of data center energy efficiency, resulting in optimization strategies that do not accurately reflect reality. This method, based on the individual energy consumption models of IT and air conditioning equipment, comprehensively considers the coupling relationships between IT and air conditioning equipment in the data center, as well as the coupling relationships between different time periods. It proposes a data center energy efficiency optimization algorithm based on multi-time-period equipment coupling, providing theoretical support for data center energy efficiency management system-level modeling and operational optimization. Attached Figure Description

[0069] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0070] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0071] Example 1:

[0072] See Figure 1 A data center energy efficiency optimization method considering multi-time period device coupling includes the following steps:

[0073] 1) Establish an energy consumption model for data center equipment;

[0074] The data center equipment energy consumption model includes an IT equipment energy consumption model, an air conditioning equipment energy consumption model, and a convection heat transfer model.

[0075] The energy consumption models for IT equipment are shown below:

[0076] P 计算 =(P max -P idle )u+P idle (1)

[0077] f(T IT )=b1+b2T IT (2)

[0078] P IT =[(P max -P idle )u+P idle ](b1+b2T IT (3)

[0079] In the formula, max and idle represent the full-load and idle states of the IT equipment, respectively; (P max -P idle u represents dynamic power consumption; P idle P represents static power; max Represents full-load power; u∈[0,1] represents the resource utilization rate of IT equipment; P 计算 f(T) represents the computing power consumption of IT equipment. IT () represents the leakage power consumption of the IT equipment; b1 and b2 are constants; T IT It refers to the chip temperature of IT equipment; P IT This indicates the energy consumption of IT equipment.

[0080] The energy consumption model for air conditioning equipment is shown below:

[0081]

[0082] In the formula, kT S represents the ambient temperature coefficient. dc P represents the sum of the projected areas of IT equipment; aircon Energy consumption of air conditioning equipment; P IT Indicates the energy consumption of IT equipment;

[0083] Among them, the total air conditioning cooling load Q aircon As shown below:

[0084] Q aircon =Q1+Q2 (5)

[0085] The heat generated by IT equipment (Q1) and the heat generated by air conditioning equipment (Q2) are shown below:

[0086] Q1 = 97% P IT (6)

[0087] Q2 = k T S dc (7)

[0088] The coefficient of performance (COP) of the air conditioner is shown below:

[0089]

[0090] In the formula, T out The air supply temperature for air conditioning equipment.

[0091] The convective heat transfer model is shown below:

[0092] T IT =P IT R in +T in (9)

[0093] T in =T out +97% DP IT (10)

[0094] In the formula, D is the heat transfer coefficient; T out The supply air temperature for air conditioning equipment; T in R is the inlet airflow temperature of the IT equipment. in The equivalent thermal resistance for inlet convection heat transfer.

[0095] 2) Establish a data center energy efficiency optimization model based on multi-time period device coupling;

[0096] The objective function of the data center energy efficiency optimization model based on multi-time period device coupling is as follows:

[0097]

[0098] In the formula, FPtotal (t)Δt represents the operating cost of the data center; P total (t) represents the total power of the data center during time period t; F is the operating cost coefficient; I i (t)[1-I i (t-1)]S i This refers to the start-up and shutdown costs of IT equipment; t represents the current time period; T is the total number of time periods; Δt is the time interval; I i (t) represents the operating state of the i-th IT device during time period t, where 0 indicates it is stopped and 1 indicates it is powered on or in standby mode; S i It is the start-up and shutdown cost coefficient.

[0099] The constraints of the data center energy efficiency optimization model based on multi-time period equipment coupling include computing power demand constraints, air conditioning supply air temperature range constraints, chip temperature upper limit constraints, inlet airflow temperature range constraints, IT equipment ramp-up constraints, and IT equipment downtime constraints.

[0100] The computing power requirement constraints are as follows:

[0101]

[0102] In the formula, u i It is the total computing power of the i-th IT device; u total This refers to the computing power currently provided by the data center; D IT This represents the computing power requirement; N is the number of IT devices.

[0103] The air conditioning supply air temperature range constraints are as follows:

[0104] T out,min ≤T out (t)≤T out,max (13)

[0105] In the formula, T out,max and T out,min These are the upper and lower limits of the air conditioner's air supply temperature; T out (t) is the air conditioner supply temperature at time t;

[0106] The upper limit constraint of chip temperature is as follows:

[0107] T IT (t)≤T IT,max (14)

[0108] In the formula, T IT,max This refers to the upper temperature limit for the normal operation of chips in IT equipment.

[0109] The inlet airflow temperature range constraints are as follows:

[0110] 18℃≤T in≤27℃ (15)

[0111] The ramp-up constraints for IT equipment are as follows:

[0112]

[0113] In the formula, R represents the power consumption of the i-th IT device during time period t when the device is powered on; IT This represents the gradient coefficient.

[0114] 3) Solve the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period equipment coupling to obtain the computing power allocation and start-up / shutdown scheme of data center IT equipment, as well as the air supply temperature of air conditioners.

[0115] The steps for solving the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period equipment coupling include:

[0116] 3.1) By simplifying the IT equipment accordingly, we get:

[0117]

[0118]

[0119] R IT,eq,i =n i ·R IT,i (19)

[0120] In the formula, u eq,i P IT,eq,i and R IT,eq,i These are the equivalent IT equipment computing power, equivalent IT equipment power consumption, and equivalent IT equipment ramp rate; u j Let n be the total computing power of the j-th IT device; i P represents the total number of IT devices. IT,j R represents the power consumption of the j-th IT device. IT,i For IT equipment ramp-up coefficient;

[0121] 3.2) Equivalent IT equipment computing power u eq,i Equivalent IT equipment power consumption P IT,eq,i And equivalent IT equipment ramp factor R IT,eq,i Substituting these into the data center equipment energy consumption model, we obtain the equivalent IT equipment energy consumption model, the equivalent air conditioning equipment energy consumption model, and the equivalent convection heat transfer model.

[0122] 3.3) Based on step 3.2), update the data center energy efficiency optimization model based on multi-time period device coupling;

[0123] 3.4) The look-ahead decoupling algorithm is used to decompose the inter-time coupling constraints, resulting in a simplified data center energy efficiency optimization model; the decomposed inter-time coupling constraints are shown below:

[0124]

[0125] N f ={i|N i ≤m}, 0<m<t (21)

[0126] N s ={i|N i >m}, 0 < m < t (22)

[0127]

[0128]

[0129] In the formula, N i Let m be the minimum number of time intervals required for the power of IT device i to increase from the lower limit to the upper limit; m is any integer; N f It is a rapidly ascending set; N s It is a slowly ascending set; It is the maximum power consumption that all IT devices can increase during the m time periods from time period t-m to time period t; This is the maximum capacity that the computing power of all IT devices can increase during the specified period; D I t T P represents the total computing power requirement of the data center during time period t. IT,imax P IT,imin Δt represents the maximum and minimum power consumption of the IT equipment; Δt represents the time difference.

[0130] The decoupling conditions for the inter-time coupling constraints after decomposition include: for any m, all satisfy...

[0131] 4) Use the branch and bound method to solve the simplified data center energy efficiency optimization model, and obtain the computing power allocation and start-up / shutdown scheme of IT equipment, as well as the final optimization results of the air supply temperature of the air conditioner.

[0132] Example 2:

[0133] A data center energy efficiency optimization method considering multi-time period device coupling includes the following steps:

[0134] 1) Energy consumption modeling of main equipment in the data center

[0135] The main energy-consuming equipment in a data center is IT equipment and air conditioning equipment. The core idea of ​​the linear model of IT equipment energy consumption modeling system utilization is to divide the energy consumption of IT equipment into two parts: the power consumption generated when processing computing load (computing power consumption) and the power loss caused by leakage current of electronic components (leakage power consumption).

[0136] The computing power consumption of IT equipment is divided into dynamic power consumption and static power consumption, as follows:

[0137] P 计算 =(P max -P idle )u+P idle (1)

[0138] Where max and idle represent the full load and idle state of IT equipment, respectively. (P) max -P idle )u represents dynamic power consumption, P idle P represents static power. max This represents the full-load power. u∈[0,1] represents the resource utilization rate of IT equipment.

[0139] The leakage power consumption of IT equipment is expressed as:

[0140] f(T IT )=b1+b2T IT (2)

[0141] Where b1 and b2 are constants, typically taken as b1 = 0.75 and b2 = 0.003125; T IT This refers to the chip temperature (K) of IT equipment. IT equipment power consumption (P) IT The calculated power consumption is expressed as the power consumption after correction for leakage current:

[0142] P IT =[(P max -P idle )u+P idle ](b1+b2T IT (3)

[0143] The energy consumption modeling of air conditioning equipment uses the "power and area method," and its load mainly comes from the heat generated by IT equipment and ambient heat, as shown below:

[0144] Q aircon =Q1+Q2 (4)

[0145] Q1 = 97% P IT (5)

[0146] Q2 = k T S dc (6)

[0147] Among them, Q aircon Q1 represents the total cooling load of the air conditioning system, and k represents the heat generated by the IT equipment. T This is the ambient temperature coefficient, which is approximately proportional to the ambient temperature. In southern my country, a coefficient of around 0.18 is suitable, while in northern China, around 0.1 is suitable. For other regions, the coefficient can be determined based on actual conditions. (S) dc This represents the sum of the projected areas of IT equipment.

[0148] The energy consumption of air conditioning equipment can be expressed as:

[0149]

[0150] COP is the coefficient of performance for air conditioning, defined as the cooling load Q. aircon Power consumption P of air conditioning equipment aircon The ratio is calculated as follows:

[0151]

[0152] Air conditioning equipment absorbs heat generated by IT equipment; this phenomenon, known in thermodynamics as "convective heat transfer," is the core relationship between IT and air conditioning equipment. Convective heat transfer can be described using an equivalent thermal resistance model, as shown in the following expression:

[0153] T IT =P IT R in +T in (9)

[0154] Among them, T in R represents the inlet airflow temperature (K). in The equivalent thermal resistance (K / W) for inlet convection heat transfer.

[0155] The relationship between the inlet air temperature of IT equipment and the supply air temperature of air conditioning equipment is expressed as follows:

[0156] T in =T out +97% DP IT (10)

[0157] Among them, T out The air supply temperature is denoted by . D is the heat transfer coefficient, which depends on the layout of the equipment.

[0158] Using the above methods, the energy consumption of the main equipment in the data center is established. Then, the sum of the data center operating cost and the start-up and shutdown costs of IT equipment is used as the objective function; the decision variables include the air conditioning supply temperature T. out and the resource utilization rate u of each IT device and its start / stop variables Ii The constraints include computing power demand constraints, air conditioning supply temperature range constraints, chip temperature upper limit constraints, inlet airflow temperature range constraints, IT equipment ramp-up constraints, and IT equipment downtime constraints. A data center energy efficiency optimization model based on multi-time period equipment coupling is established.

[0159] 2) Data center energy efficiency optimization model based on multi-time period device coupling

[0160] (1) Objective function

[0161]

[0162] Among them, FP total (t)Δt represents the operating cost of the data center, P total (t) represents the total power of the data center during time period t, F is the operating cost coefficient, and I i (t)[1-I i (t-1)]S i This refers to the start-up and shutdown costs of IT equipment; t represents the current time period, T is the total number of time periods; Δt is the time interval; I i (t) represents the operating state of the i-th IT device during time period t, where 0 indicates it is stopped and 1 indicates it is powered on or in standby mode; S i It is the start-up and shutdown cost coefficient.

[0163] (2) Constraints

[0164] ① Computing power requirement constraints

[0165]

[0166] Constraint (12) limits the total computing power of the data center. Where u i It is the total computing power of the i-th IT device, u total This refers to the computing power currently provided by the data center, D. IT It's a computing power requirement.

[0167] ② Air conditioning supply air temperature range constraints

[0168] T out,min ≤T out (t)≤T out,max (13)

[0169] Constraint (13) limits the supply air temperature of the air conditioning equipment to its operating range. out,max and T out,min These are its upper and lower limits, and their values ​​depend on the air conditioner's operating parameters.

[0170] ③ Upper limit constraint of chip temperature

[0171] T IT (t)≤TIT,max (14)

[0172] Constraint (14) limits the upper temperature limit T for the normal operation of the chip in the IT equipment. IT,max The temperature is usually 80℃.

[0173] ④ Inlet airflow temperature range constraint

[0174] 18℃≤T in ≤27℃ (15)

[0175] Constraint (15) refers to the range of inlet airflow constraints specified in the ASHRAE (Institute of Refrigeration and Air Conditioning Engineers) guidelines, which is an important reference for data center design recognized internationally.

[0176] ⑤ IT equipment ramp-up constraints

[0177]

[0178] Constraint (16) limits the ramp-up rate of IT equipment. Among them... R represents the power consumption of the i-th IT device during time period t when the device is powered on; IT This represents the gradient coefficient.

[0179] 3) Data center energy efficiency optimization algorithm based on multi-time period device coupling

[0180] This invention first establishes energy consumption models (1)-(8) and convection heat transfer models (9)-(10) for the main equipment of the data center in step 1. In step 2, it establishes a data center energy efficiency optimization model (11)-(16) based on multi-time period equipment coupling with the total cost of the data center as the objective function. Finally, it proposes a data center energy efficiency optimization algorithm based on multi-time period equipment coupling. The specific steps are as follows:

[0181] 3.1) First, the IT equipment aggregation algorithm of equations (17)-(19) is used to simplify all IT equipment, reduce the integer variables in the model calculation, and calculate the simplified equivalent parameter u. eq,i P IT,eq,i and R IT,eq,i :

[0182]

[0183]

[0184] R IT,eq,i =n i ·R IT,i (19)

[0185] Where u eq,i P IT,eq,i and RIT,eq,i These are the equivalent IT equipment computing power, equivalent IT equipment power consumption, and equivalent IT equipment ramp-up coefficient, respectively.

[0186] 3.2) Substitute the above equivalent parameters and environmental airflow parameters into equations (1)-(10) to obtain the energy consumption model and convective heat transfer model of the equivalent IT and air conditioning equipment.

[0187] 3.3) Based on the model in step 3.2), establish a multi-period energy efficiency optimization model for the data center. Utilize the look-ahead decoupling algorithm (20)-(24) to decompose the inter-period coupling constraints, reducing their frequency of action and accelerating the solution rate.

[0188]

[0189] N f ={i|N i ≤m}, 0<m<t (21)

[0190] N s ={i|N i >m}, 0 < m < t (22)

[0191]

[0192]

[0193] Where N i Let m be the minimum number of time intervals required for the power of IT device i to increase from the lower limit to the upper limit; m is any integer; N f It is a rapidly ascending set, N s It is a slowly ascending set; It is the maximum power consumption that all IT devices can increase during the m time periods from time period t-m to time period t; This is the maximum capacity that the computing power of all IT devices can increase during this period. Let m be the total computing power demand of the data center during time period t. If equation (24) holds true at least once for any m, it means that the increase in total computing power demand during those m time periods exceeds the threshold. Decoupling is not possible at this point. For any m, equation (24) takes the less than or equal to sign, then... The change in total computing power demand is greater than or equal to the change in total computing power demand. At this point, for time period t, the time period coupling constraint has no effect; this is the decoupling condition.

[0194] 4) Finally, the branch and bound method is used to solve the simplified model, and the computing power allocation and start-up / shutdown scheme of IT equipment, as well as the final optimization results of the air supply temperature of the air conditioner are obtained.

Claims

1. A data center energy efficiency optimization method considering multi-time period device coupling, characterized in that, Includes the following steps: Step 1) Establish the energy consumption model for the data center equipment; Step 2) Establish a data center energy efficiency optimization model based on multi-time period device coupling; Step 3) Solve the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period equipment coupling to obtain the computing power allocation and start-up / shutdown scheme of data center IT equipment, as well as the air supply temperature of air conditioners; The data center equipment energy consumption model includes an IT equipment energy consumption model, an air conditioning equipment energy consumption model, and a convection heat transfer model. The energy consumption models for IT equipment are shown below: (1) (2) (3) In the formula, max and idle represent the full load and idle states of the IT equipment, respectively; (P max -P idle u represents dynamic power consumption; P idle P represents static power; max Represents full-load power; u∈[0,1] represents the resource utilization rate of IT equipment; P 计算 f(T) represents the computing power consumption of IT equipment. IT () represents the leakage power consumption of the IT equipment; b1 and b2 are constants; T IT It refers to the chip temperature of IT equipment; P IT Indicates the energy consumption of IT equipment; The energy consumption model for air conditioning equipment is shown below: (4) In the formula, k T S represents the ambient temperature coefficient. dc P represents the sum of the projected areas of IT equipment; aircon Energy consumption of air conditioning equipment; P IT Indicates the energy consumption of IT equipment; Wherein, the total air conditioning cooling load Q aircon As shown below: (5) The heat generated by IT equipment (Q1) and the heat generated by air conditioning equipment (Q2) are shown below: (6) (7) The coefficient of performance (COP) of the air conditioner is shown below: (8) In the formula, T out is the supply air temperature of the air conditioning equipment; The convective heat transfer model is shown below: (9) (10) In the formula, D is the heat transfer coefficient; T out The supply air temperature for air conditioning equipment; T in R is the inlet airflow temperature of the IT equipment. in The equivalent thermal resistance for inlet convection heat transfer; The objective function of the data center energy efficiency optimization model based on multi-time period device coupling is as follows: (11) In the formula, FP total (t)Δt represents the operating cost of the data center; P total (t) represents the total power of the data center during time period t; F is the operating cost coefficient; I i (t)[1-I i (t-1)]S i This refers to the start-up and shutdown costs of IT equipment; t represents the current time period; T is the total number of time periods; Δt is the time interval; I i (t) represents the operating state of the i-th IT device during time period t, where 0 indicates it is stopped and 1 indicates it is powered on or in standby mode; S i It is the start-up and shutdown cost coefficient; The constraints of the data center energy efficiency optimization model based on multi-time period equipment coupling include computing power demand constraints, air conditioning supply air temperature range constraints, chip temperature upper limit constraints, inlet airflow temperature range constraints, IT equipment ramp-up constraints, and IT equipment downtime constraints. The computing power requirement constraints are as follows: (12) In the formula, u i It is the total computing power of the i-th IT device; u total This refers to the computing power currently provided by the data center; D IT This represents the computing power requirement; N is the number of IT devices. The air conditioning supply air temperature range constraints are as follows: (13) In the formula, T out,max and T out,min These are the upper and lower limits of the air conditioner's air supply temperature; T out (t) is the air conditioner supply temperature at time t; The upper limit constraint of chip temperature is as follows: (14) In the formula, T IT,max This refers to the upper temperature limit for the normal operation of chips in IT equipment. The inlet airflow temperature range constraints are as follows: (15) The ramp-up constraints for IT equipment are as follows: (16) In the formula, This represents the power consumption of the i-th IT device during time period t when the i-th IT device is in the startup state. For IT equipment ramp-up coefficient; The steps for solving the data center equipment energy consumption model and the data center energy efficiency optimization model based on multi-time period equipment coupling include: Step 3.1) Simplify the IT equipment accordingly to obtain: (17) (18) (19) In the formula, , and These are the equivalent IT equipment computing power, equivalent IT equipment power consumption, and equivalent IT equipment ramp rate, respectively. The total computing power of the j-th IT device; Total number of IT devices; Let J be the power consumption of the j-th IT device; For IT equipment ramp-up coefficient; Step 3.2) Equivalent IT equipment computing power Equivalent IT equipment power consumption and equivalent IT equipment ramp rate Substituting these into the data center equipment energy consumption model, we obtain the equivalent IT equipment energy consumption model, the equivalent air conditioning equipment energy consumption model, and the equivalent convection heat transfer model. Step 3.3) Based on Step 3.2), update the data center energy efficiency optimization model based on multi-time period device coupling; Step 3.4) Decompose the inter-time coupling constraints using the look-ahead decoupling algorithm to obtain a simplified data center energy efficiency optimization model; the decomposed inter-time coupling constraints are shown below: (20) (21) (22) (23) (24) In the formula, N i Let m be the minimum number of time intervals required for the power of IT device i to increase from the lower limit to the upper limit; m is any integer; N f It is a rapidly ascending set; N s It is a slowly ascending set; It is the maximum power consumption that all IT devices can increase during the m time periods from time period t-m to time period t; This is the maximum capacity that the computing power of all IT devices can increase during this period. This represents the total computing power requirement of the data center during time period t. , These are the maximum and minimum power consumption values ​​for IT equipment. For time difference; Step 3.4) Use the branch and bound method to solve the simplified data center energy efficiency optimization model to obtain the computing power allocation and start-up / shutdown scheme of IT equipment, as well as the final optimization results of the air supply temperature of the air conditioner.

2. The method for data center energy efficiency optimization considering multi-time period device coupling of claim 1, wherein, The decoupling condition of the inter-period coupling constraint after decomposition includes: for any m, both satisfy .

Images