Energy-saving method for refrigerating system of data center

Abstract

The invention relates to the field of data center energy saving, and discloses a data center refrigeration system energy saving method which comprises the following steps: 1) data monitoring and preprocessing; 2) proposing a data center refrigeration system heat exchange model; (3) according to the position and the load of the data center machine room server, CFD simulation software is used for calculating the cooling load needed by the precise air conditioner for enabling the cabinet to have the safe temperature, and an AI algorithm is used for establishing an intelligent prediction model; and 4) debugging to obtain unknown parameters of the heat transfer model, determining a critical outdoor critical wet bulb temperature TL which can completely depend on a natural cold source and a critical outdoor critical wet bulb temperature TH which can completely get rid of the natural cold source and is obtained after the working condition of the data center is stable, and establishing a time-temperature grid coordinate system in combination with a local peak-valley electricity price time rule. 5) designing an equipment opening and closing strategy of each grid in the time-temperature grid coordinate system; and 6) according to the position of the data center outdoor wet bulb temperature and time in the grid coordinate system, executing the operation corresponding to the position.

Description

Technical Field

[0001] The present invention relates to the field of data center energy saving, and in particular to an energy saving method for a data center refrigeration system. Background Art

[0002] Data centers are large facilities for storing electronic equipment. Their operation and maintenance consume a lot of electricity and generate a lot of heat. They are mainly composed of IT equipment and refrigeration systems. The refrigeration system of a data center mainly includes chillers, cold storage tanks, precision air conditioners, chilled water pumps, cooling water pumps, cooling towers, plate heat exchangers, etc. The chillers are composed of evaporators, condensers, and compressors. Precision air conditioners include fan coils and fans.

[0003] The working process of central air conditioning is a process of continuous heat exchange, which includes multiple circulation systems such as air circulation, chilled water circulation, refrigeration circulation, cooling water circulation, outdoor air circulation, etc. Under the action of the air handler, the indoor air circulation brings the heat in the house into the chilled water circulation, and the high-temperature indoor air and low-temperature chilled water exchange heat in the fan coil unit, and the cooled air enters the room. In the chilled water cycle, high-temperature chilled water enters the evaporator driven by the chilled water pump and releases heat to the refrigerant in the evaporator. The cooled chilled water exchanges heat with the indoor air again under the drive of the chilled water pump, and at the same time charges the cold storage tank with cold. In the refrigeration cycle, the low-pressure refrigerant evaporates in the evaporator and absorbs heat from the chilled water. The compressor compresses the low-pressure gaseous refrigerant into liquid refrigerant and drives the refrigerant into the condenser to exchange heat with the cooling water. The low-temperature refrigerant enters the evaporator driven by the compressor to exchange heat with the chilled water. In the cooling water cycle, the high-temperature cooling water that has absorbed the refrigerant enters the cooling tower and the plate heat exchanger. The cooling water exchanges heat with the outdoor air in the cooling tower and the plate heat exchanger. The low-temperature cooling water that has completed the heat exchange exchanges heat with the refrigerant in the condenser again under the drive of the cooling water pump. In the outdoor air circulation, the fan drives the air in the duct, discharges the hot air that has absorbed the heat of the cooling water into the atmosphere, and injects new air into the central air conditioning system (cooling water circulation). After the central air conditioning goes through the above circulation, it completes the cooling of the indoor air. The above circulation is repeated during the working process of the central air conditioning. Summary of the invention

[0004] With regard to existing cooling system facilities, most data centers currently simply charge the cold storage tank at night and discharge the cold storage tank during the day. This behavior may cause redundancy in cooling capacity and waste resources. The purpose of this invention is to establish a heat transfer model of the cooling system itself, combined with the outdoor wet bulb temperature of the data center, and a deep learning algorithm to implement a device control scheme for the cooling system.

[0005] Technical solution: To achieve the above purpose, the present invention provides an energy-saving method for a data center refrigeration system, comprising the following steps:

[0006] S1, environmental data collection and preprocessing, and transmitting the preprocessed stable values ​​to the data center monitoring room;

[0007] S2, build a heat exchange model for the data center cooling system;

[0008] S3. According to the location and load of the server in the data center computer room, use CFD simulation software to determine the cooling load required by the precision air conditioner, and use AI algorithm to establish an intelligent prediction model;

[0009] S4. Debug the unknown parameters of the heat transfer model and determine the outdoor critical wet-bulb temperature T that can be completely dependent on natural cooling sources after the data center operating conditions are stable. L The outdoor critical wet bulb temperature T H , establish a time-temperature grid coordinate system in combination with local peak and valley electricity prices;

[0010] S5. Propose a strategy for opening and closing each device for each grid in the time-temperature grid coordinate system;

[0011] S6. Execute the operation corresponding to the position according to the grid coordinate system position of the outdoor wet-bulb temperature and time of the data center.

[0012] Specifically: Step S1 includes:

[0013] S1.1. Install temperature sensors, humidity sensors, and power equipment sensors near the cooling towers and plate heat exchangers in the data center, as well as at the weather station;

[0014] S1.2. Use methods including data cleaning to process the large amount of collected environmental data, then detect and delete missing values, use interpolation to fill missing values, detect and eliminate outliers, and transmit stable values ​​to the data center monitoring room as a temperature reference.

[0015] Specific:

[0016] The heat exchange model of the data center refrigeration system in step S2:

[0017] When the outdoor wet-bulb temperature t A satisfy: When cooling, it completely relies on natural cooling sources;

[0018] When the outdoor wet-bulb temperature t A satisfy: When using, water chillers and plate heat exchangers are used together, and natural cooling sources are used in part;

[0019] When the outdoor wet-bulb temperature t A satisfy: When the temperature is high, the strategy of using the chiller for cooling is adopted; among them, Q AIR Indicates the cooling capacity of precision air conditioner, T in represents the cooling water inlet temperature, a represents the precision air conditioning heating coefficient, t A Represents the outdoor wet bulb temperature, V cool represents the cooling water flow rate, c p represents the specific heat capacity of water, ρ represents the density of water, ΔT E It represents the difference between the outdoor air wet bulb temperature and the cooling tower outlet water temperature, where Ω refers to To simplify the expression, U represents the heat transfer coefficient of the plate heat exchanger, A represents the heat transfer area, COP represents the performance coefficient of the data center, Δt p Represents the mean temperature difference.

[0020] Specifically: Step S3 includes:

[0021] S3.1 Input the characteristic parameters of each cabinet into the CFD software 6sigmaDC according to the actual layout of the cabinets in the data center computer room. The characteristic parameters include load, quantity, and precision air conditioning position;

[0022] S3.2 Set the precision air conditioner in the software to automatically adjust the outlet temperature so that all cabinets can operate at a safe temperature; after the CFD software simulation settlement is completed, read the outlet temperature T corresponding to the precision air conditioner at the final stability out and return air temperature T in , so the cooling capacity required by the precision air conditioner can be calculated by the cooling capacity formula of the precision air conditioner; the cooling capacity formula of the precision air conditioner is:

[0023] Q AIR =V air ×ρ air ×c air ×(T air,in -T air,out )

[0024] Among them, Q AIR represents the cooling capacity of precision air conditioning, ρ air represents the air density, c air represents the specific heat capacity of water, T air,in Indicates the return air temperature, T air,out Indicates the outlet air temperature;

[0025] S3.3 By inputting the historical data of the room load and the location data into the CFD model, the required inlet and outlet air temperatures of the precision air conditioner are obtained in turn, and the required cooling capacity is calculated in turn;

[0026] S3.4 organizes the data obtained in S3.3, including the cabinet load, location, and precision air conditioning inlet and outlet air temperatures; establishes a data center precision air conditioning inlet and outlet air temperature prediction model through the deep learning method of graph neural network; wherein the server load of the cabinet is used as the input parameter in the neural network, and the outlet air temperature and return air temperature of the precision air conditioning are used as the output parameters;

[0027] S3.5 Input the trained data center precision air conditioning inlet and outlet air temperature prediction model into the precision air conditioning control system. The system calculates the precision air conditioning inlet and outlet air temperature T according to the characteristic parameters of the monitored data center computer room. air,in and T air,out , combined with the cooling capacity formula of the precision air conditioner, the required cooling capacity Q is obtained AIR .

[0028] Specifically: Step S4 includes:

[0029] S4.1 Based on the monitoring of sensor temperature, the COP value of the data center and the temperature difference ΔT of the cooling water inlet and outlet of the plate heat exchanger are confirmed in real time P , the difference between the outdoor air wet bulb temperature and the cooling tower outlet water temperature ΔT E , and the specific values ​​of each parameter, including Q AIR 、T in , a, t A 、V cool 、c p ,ρ,Ω,U,A,COP,Δt p ;

[0030] S4.2 Based on the heat transfer formula obtained in S2 and combined with the ASHARE standard, the outdoor critical wet bulb temperature is obtained and The specific value of

[0031] S4.3 Binding T L and T H The time-temperature grid coordinate system is formed with the peak and valley electricity prices.

[0032] The step S5 specifically comprises: L and T H Combined with the peak and valley electricity price schedule of the data center, and based on this, operation and shutdown strategies for chillers, natural cooling sources, and cold storage tanks are proposed according to different temperatures in different time periods.

[0033] More specific:

[0034] Determine t0~t67 time points and divide 24 hours a day into eight sections, among which t0 is the cold storage tank charging warning time point; the peak electricity price time period t3-t2 and t5-t4 is the J zone; the peak electricity price time period t2-t1 and t6-t5 is the G zone; the valley electricity price time period t1-t6 and t4-t3 is the D zone; the temperature is determined by T L and T H Divided into three areas, T 1区 >T H , T H >T 2区 >T L , T L >T 3区 >0, where: T 1区 、T 2区 、T 3区 Respectively represent the temperature of zone 1, zone 2, and zone 3;

[0035] The specific operation opening and closing strategies of the chiller, natural cooling source and cold storage tank are:

[0036] ① The cold storage tank must be filled with cold water at the beginning of time t0;

[0037] ② In the interval of t1 to t2, if the temperature is lower than T L , then use natural cooling source for independent cooling; if the temperature is T L and T H If the temperature is higher than T H , then the chiller is supplied with cold, and the cold storage tank is not charged with cold during this period;

[0038] ③ In the interval of t2 to t3, J3 uses natural cold source to independently supply cooling according to formulas B1 and B2, J2 and J1 open the cold storage tank to cool, and the chiller is turned off;

[0039] ④ In the interval of t3-t4, D6 uses natural cold source to independently supply cooling according to formulas B1 and B2, D5 closes the cold storage tank to cool, and uses the chiller to cooperate with the natural cold source to supply cooling according to formulas A1 and A2, during which the cold storage tank is charged with cold; D6 closes the cold storage tank to cool, and uses the chiller to cooperate with the natural cold source to supply cooling according to formulas A1 and A2, during which the cold storage tank is charged with cold;

[0040] ⑤ In the interval of t4-t5, J6 uses natural cold source to independently supply cooling according to formulas B1 and B2, J4 and J5 open the cold storage tank to cool, and the chiller is turned off;

[0041] ⑥ In the interval of t5 to t6, G6 uses natural cold source to independently provide cooling according to formulas B1 and B2. G5 determines whether the cold capacity of the cold storage tank is exhausted. If it is exhausted, the chiller cooperates with the natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is not charged during this period; if it is not exhausted, the cold storage tank is discharged; G4 determines whether the cold capacity of the cold storage tank is exhausted. If it is exhausted, the chiller cooperates with the natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is not charged during this period; if it is not exhausted, the cold storage tank is discharged;

[0042] ⑦ In the interval of t6 to t1, D3 uses natural cold source to independently provide cooling according to formulas B1 and B2. When t0 is reached, it is determined whether the cold storage tank is full. If it is full, the cold storage tank is charged with cold water by the chiller. D2 determines whether the cold storage tank is exhausted. If it is not exhausted, the cold storage tank is used to cool down, and it is determined whether t0 has been reached. If it has been reached, the chiller is used in conjunction with natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is charged with cold during this period. If it is exhausted, the chiller is used in conjunction with natural cold source to provide cooling according to formulas A1 and A2, and no The cold storage tank is charged with cold, and it is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold. D1 determines whether the cold storage tank is exhausted. If not, the cold storage tank is used to cool, and it is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold water in cooperation with the natural cold source according to formulas A1 and A2. If exhausted, the cold storage tank is not charged with cold water in cooperation with the natural cold source according to formulas A1 and A2. It is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold water in cooperation with the natural cold source. Wherein:

[0043] The formula A1 is: chiller +Q plate ≥(1+a)Q AIR ;

[0044] The formula A2 is: cooltower =Q chiller +Q plate +P chiller ;

[0045] The formula B1 is: plate ≥(1+a)Q AIR ;

[0046] The formula B2 is: cooltower =Q plate .

[0047] Where: Q AIR :Precision air conditioning cooling capacity, unit kW; Q chiller : Cooling capacity of chiller, unit kW; Q cooltower : Cooling tower heat exchange capacity, unit kW; Q plate: Plate heat exchanger heat transfer capacity, unit kW; a is the precision air conditioning heating coefficient between 0.06 and 0.07.

[0048] The step S6 specifically includes: collecting data in combination with the outdoor temperature, determining the position of the temperature and time in the table, and operating the device as a strategy based on the corresponding requirements of the table position.

[0049] Beneficial effects:

[0050] Compared with the conventional unchanged power consumption mode of equipment in data centers, this solution dynamically opens and closes each device in the data center's cooling system by referring to the peak and valley electricity price policy and outdoor wet-bulb temperature, thereby greatly reducing power consumption and costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of the data center energy saving strategy proposed by the present invention.

[0052] Figure 2 The critical temperature T that is completely dependent on natural cooling source proposed by the present invention is L The critical temperature T that is completely separated from the natural cooling source in this working condition H A time-temperature grid coordinate system that combines the two with peak and valley electricity prices. DETAILED DESCRIPTION

[0053] The technical solutions in the examples of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0054] See also Figure 1-2 The present invention is an energy-saving method for a data center refrigeration system. Specifically, the method comprises the following steps:

[0055] To achieve the above object, the present invention provides the following technical solution: an energy-saving method for a data center refrigeration system, comprising the following steps:

[0056] S1. Data monitoring and preprocessing,

[0057] S2. Propose a heat exchange model for data center cooling system;

[0058] S3. Based on the location and load of the servers in the data center computer room, use CFD simulation software to calculate the cooling load required for the cabinet to work safely, and use AI algorithm to establish an intelligent prediction model for the cooling capacity of precision air conditioners;

[0059] S4. Debug the unknown parameters of the heat exchange model and determine the outdoor critical wet-bulb temperature T of the cooling system that can fully rely on natural cooling sources after the data center operating conditions are stable. L The outdoor critical wet bulb temperature T H, and establish a time-temperature coordinate system in combination with peak and valley electricity prices, and divide the grid.

[0060] S5. Based on the monitoring data and combined with the peak and valley electricity prices, propose the refrigeration system operation strategy for each grid in the coordinate system;

[0061] S6. Combine the data collected with the outdoor temperature to analyze which grid the current situation is in and operate the equipment according to the strategy.

[0062] Furthermore, S1 includes:

[0063] S1.1. Install temperature sensors and power equipment sensors near the cooling towers and plate heat exchangers in the data center, as well as at the weather station. Specifically, install temperature sensors at the chilled water inlet and outlet of the chiller, the chilled water net outlet of the plate heat exchanger, and the net outlet of the cooling tower.

[0064] S1.2. Use data cleaning to process the large amount of collected data first, then detect and delete missing values, use interpolation to fill missing values, then detect and eliminate outliers by setting reasonable thresholds and using statistical methods, and transmit stable values ​​to the data center monitoring room as a temperature reference

[0065] Furthermore, the step S2 comprises:

[0066] S2.1. First determine the total cooling capacity (cooling capacity) formula

[0067] W=Q×t

[0068] W: total cooling capacity (cooling capacity), kWh; Q: cooling capacity, kW; t: time, h

[0069] S2.2. Heat transfer formula of cooling tower

[0070] Q cooltower =V cool ×ρ×c p ×ΔT

[0071] Q cooltower : Cooling tower heat exchange, kW; V cool :Cooling water flow, m3 / s; Water density: ρ, kg / m 3 ;

[0072] c p : specific heat capacity of water; ΔT: temperature difference between inlet and outlet of cooling water, ℃

[0073] ΔT=T in -T out

[0074] T in is the cooling water inlet temperature, T outCooling water outlet temperature

[0075] S2.3, Plate heat exchanger heat transfer formula

[0076] Q plate =U×A×Δt p

[0077] Q plate :Heat transfer capacity of plate heat exchanger, kW U:Total heat transfer coefficient of plate heat exchanger A:Heat transfer area, m3

[0078] Δt p :Average temperature difference,℃

[0079] S2.4, Chiller cooling capacity formula

[0080]

[0081] Q chiller :Chiller cooling capacity, kW t chill,in : Chilled water inlet temperature, ℃t chill,out :Chilled water outlet temperature,℃,Δt chill =t chill,in -t chill,out : Chilled water temperature difference, ℃, C: Chilled water specific heat capacity

[0082] S2.6, total cold storage capacity of cold storage tank W TANK =cρV TANK (t chill,in -t chill,out )

[0083] V TANK : Cold storage tank volume, m 3

[0084] S2.7 Precision air conditioning cooling capacity formula

[0085] Q AIR =V air ×ρ air ×c air ×(T air,in -T air,out )

[0086] Q AIR :Precision air conditioning cooling capacity, kWρ air : Air density c air ,kg / m 3 : Specific heat capacity of water

[0087] T air,in : Air conditioning return air temperature, °C T air,out :Air conditioning outlet temperature, ℃

[0088] S2.8 Cold storage tank cold storage time

[0089]

[0090] t TANK,STO :Cold storage time,h

[0091] S2.9 Cooling time of cold storage tank

[0092]

[0093] t TANK,DIS : Cooling time of cold storage tank, hη TANK : Cold storage efficiency p: Cold storage tank volume ratio

[0094] Assuming that the cooling capacity of the cold storage tank remains unchanged, the cooling capacity of the cold storage tank is

[0095]

[0096] S2.10 links the heat exchange between precision air conditioning cooling capacity, chillers, cold storage tanks, cooling towers, and plate heat exchangers:

[0097] (1) In general, the data center is in a state of cooling and heat exchange with natural cooling sources and chillers

[0098] Formula A1 is:

[0099] Q chiller +Q plate ≥(1+a)Q AIR

[0100] Formula A2 is:

[0101] Q cooltower =Q chiller +Q plate +P chiller

[0102] Where a is the precision air conditioning heating coefficient, which is between 0.06 and 0.07.

[0103] according to Where COP is the coefficient of performance, which can be obtained through monitoring, thus:

[0104]

[0105] Q in the above formula chiller Use Q cooltower replace:

[0106]

[0107]

[0108] make Then we can get:

[0109] COP·V cool ·ρ·c p ·ΔT+U·A·Δt p ≥Ω

[0110]

[0111] Assume that the difference between the outdoor air wet bulb temperature and the cooling tower outlet water temperature is ΔT E , the outdoor wet bulb temperature is t A , then we can get:

[0112] t A +ΔT E =T out

[0113]

[0114] (2) When the temperature is so low that it completely relies on natural cooling sources,

[0115] Formula B1 is:

[0116] Q plate ≥(1+a)Q AIR

[0117] Formula B2 is:

[0118] Q cooltower =Q plate

[0119] The calculation yields:

[0120] V cool ·ρ·c p ·ΔT≥(1+A)·Q AIR

[0121]

[0122] In summary, when the outdoor wet-bulb temperature When the outdoor wet-bulb temperature is When the outdoor wet bulb temperature is When cooling, the strategy of using chillers completely is adopted.

[0123] The schematic diagram and units of the parameters involved are shown in Table 1:

[0124] Table 1: Parameter meaning and unit

[0125]

[0126] Furthermore, the step S3 includes:

[0127] S3.1 According to the actual layout of cabinets in the data center computer room, the characteristic parameters such as the load, quantity, and location of precision air conditioners of each cabinet can be input into the CFD software 6sigmaDC.

[0128] S3.2 Set the precision air conditioner in the software to automatically adjust the outlet temperature so that all cabinets can operate at a safe temperature. After the CFD software simulation settlement is completed, the outlet temperature T corresponding to the precision air conditioner at the final stability can be read. out and return air temperature T in , so the cooling capacity required by the precision air conditioner can be calculated using the S2.7 formula.

[0129] S3.3 By inputting the historical data of the room load and the location data into the CFD model in batches, the required precision air conditioning inlet and outlet air temperatures can be obtained in turn, and the required cooling capacity Q can be calculated in turn. AIR .

[0130] S3.4 organizes the data obtained in S3.3, including cabinet load, position, and air inlet and outlet temperature of precision air conditioners. It can be trained through the graph neural network regression algorithm, and the cabinet range affected by each precision air conditioner can be judged by the distance relationship between the precision air conditioner and the cabinet, so that all cabinets within the cabinet range can form a graph model. In the graph model, each cabinet is numbered from 0, and if there are k servers, they are numbered up to k-1. Point data Each cabinet has a total of n servers from bottom to top, as a set of n-dimensional data, each dimension of data is the steady-state load L corresponding to each server. If there is no server stored in the corresponding position, the corresponding dimension data is set to 0. Make edges according to the positional relationship between cabinets, and select each cabinet as a cabinet with connections with itself, adjacent cabinets, opposite cabinets, and diagonal cabinets.

[0131] The format of point data is $D={L_1,L_2,L_3,L_4,…,L_n}.

[0132] The edge data format is [[0,0], [0,1], [0,2], …, [k-2,k-1], [k-1,k-1]]

[0133] The characteristic of this algorithm is that it takes into account the influence of cabinet location on heat transfer.

[0134] The algorithm output is set to 2D data, which is the outlet and return air temperature of each precision air conditioner.

[0135] After the algorithm training is completed, a prediction model for the inlet and outlet air temperatures of the data center precision air conditioning is obtained.

[0136] S3.5 inputs the trained prediction model into the precision air conditioning control system. The system obtains the inlet and outlet air temperatures of the precision air conditioning based on the characteristic parameters of the monitored data center room, and calculates the required cooling capacity Q through the formula AIR .

[0137] Furthermore, the step S4 comprises:

[0138] S4.1 Based on the monitoring of sensor temperature, the COP value of the data center and the temperature difference ΔT of the cooling water inlet and outlet of the plate heat exchanger are confirmed in real time P , the temperature difference between the cooling tower outlet water temperature and the ambient wet bulb temperature ΔT E , and the specific values ​​of each parameter in Table 1 above.

[0139] S4.2 Based on the heat transfer formula obtained in S2 and combined with the ASHARE standard, it is concluded that when the outdoor critical wet bulb temperature and The specific value of .

[0140] S4.3 Substitute the T obtained in S4.4 L and T H Combined with the peak and valley electricity price schedule of the data center. Figure 2 The time-temperature grid coordinate system shown.

[0141] This patent selects the peak and valley electricity price periods of Shanghai's industrial electricity consumption as a reference object. Table 2 shows the peak and valley periods of Shanghai's industrial electricity consumption. If there are differences in other cities, there is no connection between the other time periods except the low point at night and the peak point thereafter. If there are differences in the peak and valley time periods in different cities, the amount can be increased or decreased accordingly.

[0142] Table 2 Peak and valley periods of industrial electricity consumption in Shanghai

[0143] Lows peak peak Lows peak peak Lows

[0144] Furthermore, the step 5 comprises:

[0145] Determine t0~t67 time points and divide 24 hours a day into eight sections, among which t0 is the cold storage tank charging warning time point; the peak electricity price time period t3-t2 and t5-t4 is the J zone; the peak electricity price time period t2-t1 and t6-t5 is the G zone; the valley electricity price time period t1-t6 and t4-t3 is the D zone; the temperature is determined by T L and T H Divided into three areas, T 1区 >T H , T H >T 2区 >T L , TL >T 3区 >0;

[0146] The specific operation opening and closing strategies of the chiller, natural cooling source and cold storage tank are:

[0147] ① The cold storage tank must be filled with cold water at the beginning of time t0;

[0148] ② In the interval of t1 to t2, if the temperature is lower than T L , then use natural cooling source for independent cooling; if the temperature is T L and T H If the temperature is higher than T H , then the chiller is supplied with cold, and the cold storage tank is not charged with cold during this period;

[0149] ③ In the interval of t2 to t3, J3 uses natural cold source to independently supply cooling according to formulas B1 and B2, J2 and J1 open the cold storage tank to cool, and the chiller is turned off;

[0150] ④ In the interval of t3-t4, D6 uses natural cold source to independently supply cooling according to formulas B1 and B2, D5 closes the cold storage tank to cool, and uses the chiller to cooperate with the natural cold source to supply cooling according to formulas A1 and A2, during which the cold storage tank is charged with cold; D6 closes the cold storage tank to cool, and uses the chiller to cooperate with the natural cold source to supply cooling according to formulas A1 and A2, during which the cold storage tank is charged with cold;

[0151] ⑤ In the interval of t4-t5, J6 uses natural cold source to independently supply cooling according to formulas B1 and B2, J4 and J5 open the cold storage tank to cool, and the chiller is turned off;

[0152] ⑥ In the interval of t5 to t6, G6 uses natural cold source to independently provide cooling according to formulas B1 and B2. G5 determines whether the cold capacity of the cold storage tank is exhausted. If it is exhausted, the chiller cooperates with the natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is not charged during this period; if it is not exhausted, the cold storage tank is discharged; G4 determines whether the cold capacity of the cold storage tank is exhausted. If it is exhausted, the chiller cooperates with the natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is not charged during this period; if it is not exhausted, the cold storage tank is discharged;

[0153] ⑦ In the interval of t6 to t1, D3 uses natural cold source to independently provide cooling according to formulas B1 and B2. When t0 is reached, it is determined whether the cold storage tank is full. If it is full, the cold storage tank is charged with cold water by the chiller. D2 determines whether the cold storage tank is exhausted. If it is not exhausted, the cold storage tank is used to cool down, and it is determined whether t0 has been reached. If it has been reached, the chiller is used in conjunction with natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is charged with cold during this period. If it is exhausted, the chiller is used in conjunction with natural cold source to provide cooling according to formulas A1 and A2, and no The cold storage tank is charged with cold, and it is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold. D1 determines whether the cold storage tank is exhausted. If not, the cold storage tank is used to cool, and it is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold water in cooperation with the natural cold source according to formulas A1 and A2. If exhausted, the cold storage tank is not charged with cold water in cooperation with the natural cold source according to formulas A1 and A2. It is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold water in cooperation with the natural cold source. Wherein:

[0154] The formula A1 is: chiller +Q plate ≥(1+a)Q AIR ;

[0155] The formula A2 is: cooltower =Q chiller +Q plate +P chiller ;

[0156] The formula B1 is: plate ≥(1+a)Q AIR ;

[0157] The formula B2 is: cooltower =Q plate ;

[0158] Where: Q AIR : Precision air conditioning cooling capacity, kW; Q chiller : Cooling capacity of chiller, kW; Q cooltower : Cooling tower heat exchange, kW; Q plate : Heat transfer capacity of plate heat exchanger, kW; a is the heating coefficient of precision air conditioner, between 0.06 and 0.07.

[0159] Further, the step S6 includes:

[0160] For the real-time temperature and peak-valley time periods, analyze which grid number it is in and perform the corresponding operation for that grid.

[0161] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. An energy-saving method for a data center refrigeration system, characterized in that The following steps are involved: S1, environmental data collection and preprocessing, and transmitting the preprocessed stable values ​​to the data center monitoring room; S2, build a heat exchange model for the data center cooling system; S3. According to the location and load of the server in the data center computer room, use CFD simulation software to determine the cooling load required by the precision air conditioner, and use AI algorithm to establish an intelligent prediction model; S4. Debug the unknown parameters of the heat transfer model and determine the outdoor critical wet-bulb temperature T that can be completely dependent on natural cooling sources after the data center operating conditions are stable. L The outdoor critical wet bulb temperature T H , establish a time-temperature grid coordinate system in combination with local peak and valley electricity prices; S5. Propose a strategy for opening and closing each device for each grid in the time-temperature grid coordinate system; S6. Execute the operation corresponding to the position according to the grid coordinate system position of the outdoor wet-bulb temperature and time of the data center.

2. The data center refrigeration system energy saving method according to claim 1, characterized in that: The step S1 comprises: S1.

1. Install temperature sensors, humidity sensors, and power equipment sensors near the cooling towers and plate heat exchangers in the data center, as well as at the weather station; S1.

2. Use methods including data cleaning to process the large amount of collected environmental data, then detect and delete missing values, use interpolation to fill missing values, detect and eliminate outliers, and transmit stable values ​​to the data center monitoring room as a temperature reference.

3. The energy-saving method for a data center refrigeration system according to claim 1, characterized in that: The heat exchange model of the data center refrigeration system in step S2: When the outdoor wet-bulb temperature t A Satisfy: t A ≤T L When the cooling is completely dependent on natural cooling sources, When the outdoor wet-bulb temperature t A Satisfaction: T L ≤t A ≤T H When using a chiller and a plate heat exchanger, some natural cooling sources are used. When the outdoor wet-bulb temperature t A Satisfy: t A ≥T H When the cooling is complete, the strategy of using chillers for cooling is adopted; Among them, Q AIR Indicates the cooling capacity of precision air conditioner, T in represents the cooling water inlet temperature, a represents the precision air conditioning heating coefficient, t A Indicates the outdoor wet bulb temperature, V cool represents the cooling water flow rate, c p represents the specific heat capacity of water, ρ represents the density of water, ΔT E It represents the difference between the outdoor air wet bulb temperature and the cooling tower outlet water temperature, where Ω refers to To simplify the expression, U represents the heat transfer coefficient of the plate heat exchanger, A represents the heat transfer area, COP represents the performance coefficient of the data center, Δt p Represents the mean temperature difference.

4. The data center refrigeration system energy saving method according to claim 1, characterized in that: The step S3 comprises: S3.1 Input the characteristic parameters of each cabinet into the CFD software 6sigmaDC according to the actual layout of the cabinets in the data center computer room. The characteristic parameters include load, quantity, and precision air conditioning position; S3.2 Set the precision air conditioner in the software to automatically adjust the outlet temperature so that all cabinets can operate at a safe temperature; after the CFD software simulation settlement is completed, read the outlet temperature T corresponding to the precision air conditioner at the final stability out and return air temperature T in , so the cooling capacity required by the precision air conditioner can be calculated by the cooling capacity formula of the precision air conditioner; the cooling capacity formula of the precision air conditioner is: Q AIR =V air ×ρ air ×c air ×(T air,in -T air,out ) Among them, Q AIR represents the cooling capacity of precision air conditioning, ρ air represents the air density, c air represents the specific heat capacity of water, T air,in Indicates the return air temperature, T air,out Indicates the outlet air temperature; S3.3 By inputting the historical data of the room load and the location data into the CFD model, the required inlet and outlet air temperatures of the precision air conditioner are obtained in turn, and the required cooling capacity is calculated in turn; S3.4 organizes the data obtained in S3.3, including the cabinet load, location, and precision air conditioning inlet and outlet air temperatures; establishes a data center precision air conditioning inlet and outlet air temperature prediction model through the deep learning method of graph neural network; wherein the server load of the cabinet is used as the input parameter in the neural network, and the outlet air temperature and return air temperature of the precision air conditioning are used as the output parameters; S3.5 Input the trained data center precision air conditioning inlet and outlet air temperature prediction model into the precision air conditioning control system. The system calculates the precision air conditioning inlet and outlet air temperature T according to the characteristic parameters of the monitored data center computer room. air,in and T air,out , combined with the cooling capacity formula of the precision air conditioner, the required cooling capacity Q is obtained AIR .

5. The data center refrigeration system energy saving method according to claim 1, characterized in that: The step S4 comprises: S4.1 Based on the monitoring of sensor temperature, the COP value of the data center and the temperature difference ΔT of the cooling water inlet and outlet of the plate heat exchanger are confirmed in real time P , the difference between the outdoor air wet bulb temperature and the cooling tower outlet water temperature ΔT E , and the specific values ​​of each parameter, including Q AIR 、T in , a, t A 、V cool 、c p ,ρ,Ω,U,A,COP,Δt p ; S4.2 Conclusion and The specific value of S4.3 Binding T L and T H The time-temperature grid coordinate system is formed with the peak and valley electricity prices.

6. The data center refrigeration system energy saving method according to claim 1, characterized in that: The step S5 specifically comprises: L and T H Combined with the peak and valley electricity price schedule of the data center, and based on this, operation and shutdown strategies for chillers, natural cooling sources, and cold storage tanks are proposed according to different temperatures in different time periods.

7. A data center refrigeration system energy saving method according to claim 6, characterized in that: The step S5 specifically comprises: Determine t0~t67 time points and divide 24 hours a day into eight sections, among which t0 is the cold storage tank charging warning time point; the peak electricity price time period t3-t2 and t5-t4 is the J zone; the peak electricity price time period t2-t1 and t6-t5 is the G zone; the valley electricity price time period t1-t6 and t4-t3 is the D zone; the temperature is determined by T L and T H Divided into three areas, T 1区 >T H , T H >T 2区 >T L , T L >T 3区 >0, where: T 1区 、T 2区 、T 3区 Respectively represent the temperature of zone 1, zone 2, and zone 3; The specific operation opening and closing strategies of the chiller, natural cooling source and cold storage tank are: ① The cold storage tank must be filled with cold water at the beginning of time t0; ② In the interval of t1 to t2, if the temperature is lower than T L , then use natural cooling source for independent cooling; if the temperature is T L and T H If the temperature is higher than T H , then the chiller is supplied with cold, and the cold storage tank is not charged with cold during this period; ③ In the interval of t2 to t3, J3 uses natural cold source to independently supply cooling according to formulas B1 and B2, J2 and J1 open the cold storage tank to cool, and the chiller is turned off; ④ In the interval of t3-t4, D6 uses natural cold source to independently supply cooling according to formulas B1 and B2, D5 closes the cold storage tank to cool, and uses the chiller to cooperate with the natural cold source to supply cooling according to formulas A1 and A2, during which the cold storage tank is charged with cold; D6 closes the cold storage tank to cool, and uses the chiller to cooperate with the natural cold source to supply cooling according to formulas A1 and A2, during which the cold storage tank is charged with cold; ⑤ In the interval of t4-t5, J6 uses natural cold source to independently supply cooling according to formulas B1 and B2, J4 and J5 open the cold storage tank to cool, and the chiller is turned off; ⑥ In the interval of t5 to t6, G6 uses natural cold source to independently provide cooling according to formulas B1 and B2. G5 determines whether the cold capacity of the cold storage tank is exhausted. If it is exhausted, the chiller cooperates with the natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is not charged during this period; if it is not exhausted, the cold storage tank is discharged; G4 determines whether the cold capacity of the cold storage tank is exhausted. If it is exhausted, the chiller cooperates with the natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is not charged during this period; if it is not exhausted, the cold storage tank is discharged; ⑦ In the interval of t6 to t1, D3 uses natural cold source to independently provide cooling according to formulas B1 and B2. When t0 is reached, it is determined whether the cold storage tank is full. If it is full, the cold storage tank is charged with cold water by the chiller. D2 determines whether the cold storage tank is exhausted. If it is not exhausted, the cold storage tank is used to cool down, and it is determined whether t0 has been reached. If it has been reached, the chiller is used in conjunction with natural cold source to provide cooling according to formulas A1 and A2, and the cold storage tank is charged with cold during this period. If it is exhausted, the chiller is used in conjunction with natural cold source to provide cooling according to formulas A1 and A2, and no The cold storage tank is charged with cold, and it is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold. D1 determines whether the cold storage tank is exhausted. If not, the cold storage tank is used to cool, and it is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold water in cooperation with the natural cold source according to formulas A1 and A2. If exhausted, the cold storage tank is not charged with cold water in cooperation with the natural cold source according to formulas A1 and A2. It is determined whether the time t0 has been reached. If so, the cold storage tank is charged with cold water in cooperation with the natural cold source. Wherein: The formula A1 is: chiller +Q plate ≥(1+a)Q AIR ; The formula A2 is: cooltower =Q chiller +Q plate +P chiller ; The formula B1 is: plate ≥(1+a)Q AIR ; The formula B2 is: cooltower =Q plate ; Where: Q AIR :Precision air conditioning cooling capacity, unit kW; Q chiller : Cooling capacity of chiller, unit kW; Q cooltower : Cooling tower heat exchange capacity, unit kW; Q plate : Plate heat exchanger heat transfer capacity, unit kW; a is the precision air conditioning heating coefficient between 0.06 and 0.

07.

8. The data center refrigeration system energy saving method according to claim 1, characterized in that: The step S6 specifically includes: collecting data in combination with the outdoor temperature, determining the position of the temperature and time in the table, and operating the device as a strategy based on the corresponding requirements of the table position.

Images