Parameter adjusting method, medium and system for double-circulation refrigeration water system of machine room

By establishing the operating equation system and multi-objective optimization model of the dual-circulation refrigeration water system, combining the gray wolf hunting algorithm and gradient descent method, the system parameters are dynamically adjusted, and the dual-circulation refrigeration system is difficult to match the server's real-time requirements and take into account both energy consumption and performance. The goal of maximizing refrigeration performance and minimizing energy consumption has been achieved.

CN119053128APending Publication Date: 2024-11-29QINGDAO GUOSHI DATA SERVICE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411389690.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-08
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The existing dual-circulation refrigeration system is difficult to accurately match the server's real-time refrigeration needs, it is difficult to balance the system's energy consumption and refrigeration performance, and it lacks global optimization of the system's operating status.

Method used

Establish the operating equations of the dual circulation refrigeration water system in the computer room, establish the dual circulation model by fitting historical data, adopt the multi-objective cold volume allocation optimization model and a single-server cooling model, and combine the gray wolf hunting algorithm and gradient descent method to dynamically adjust the system parameters To maximize refrigeration performance and minimize energy consumption.

Benefits of technology

It accurately matches the server's real-time cooling requirements, takes into account the balance of system energy consumption and cooling performance, realizes global optimization of the system operating status, and has good adaptability and robustness to ensure the best performance under various operating conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119053128A_ABST
    Figure CN119053128A_ABST
Patent Text Reader

Abstract

The invention provides a machine room double-circulation refrigeration water system parameter adjusting method, medium and system, and belongs to the technical field of machine room double-circulation refrigeration water systems.The machine room double-circulation refrigeration water system parameter adjusting method comprises the steps that firstly, an operation equation set considering server calculation business volume and system operation parameters is established, fitting is conducted through historical data, and a double-circulation model is obtained; based on the model and real-time data, a double-layer model comprising a multi-target cooling capacity distribution optimization model and a single-server cooling model is established. A multi-objective optimization problem is solved through a grey wolf algorithm, a preliminary cooling capacity distribution scheme is obtained, and then a precise cooling capacity demand is obtained through single server model optimization. And finally, a system adjustment parameter target value is calculated and dynamically adjusted, the optimization process is circularly executed, intelligent and accurate control over the machine room refrigeration system is achieved, and the energy efficiency and stability are improved. According to the method, the medium and the system, the technical problem that in the prior art, various key parameters of the double-circulation refrigerating system are difficult to adjust intelligently is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of double-circulation cooling water systems for computer rooms, and in particular, relates to a parameter adjustment method, medium and system for double-circulation cooling water systems for computer rooms. Background Art

[0002] At present, with the rapid development of technologies such as cloud computing and big data, the scale of data centers continues to expand, and the server load continues to increase. This has brought huge challenges to the cooling system of data centers. The traditional single-cycle refrigeration system can no longer meet the growing cooling demand, and its energy consumption is too high. Therefore, data centers have begun to generally adopt a dual-cycle refrigeration system, that is, a computer room refrigeration system that combines a refrigeration cycle and a cooling cycle, to improve the overall cooling performance and energy efficiency.

[0003] The core of the dual-cycle refrigeration system is to provide cooling capacity through the refrigeration cycle and discharge heat to the external environment through the cooling cycle. This system structure can effectively improve the refrigeration efficiency of the refrigeration system and reduce overall energy consumption. But at the same time, due to the complexity of the system structure, how to reasonably allocate the refrigeration load between the refrigeration cycle and the cooling cycle has become a key factor restricting the performance of the dual-cycle refrigeration system. At present, the industry generally adopts an empirical manual adjustment method, that is, adjusting the system parameters based on the experience of engineers. There are some problems with this method:

[0004] 1) It is difficult to accurately match the real-time cooling needs of the server. The server computing workload changes dramatically, and manual adjustment is difficult to respond to such changes quickly, which can easily cause the server temperature to exceed the range.

[0005] 2) It is difficult to strike a balance between system energy consumption and refrigeration performance. Manual adjustment usually focuses more on meeting refrigeration needs and ignores the optimization of energy consumption, resulting in high system energy consumption.

[0006] 3) Lack of global optimization of system operation status. Manual adjustment is limited to the optimization of a single parameter, making it difficult to achieve comprehensive optimization of the overall system performance.

[0007] Therefore, there is an urgent need for an optimization method that can dynamically sense changes in server load and intelligently adjust the key parameters of the dual-cycle refrigeration system in order to achieve the goal of maximizing refrigeration performance and minimizing energy consumption. Summary of the invention

[0008] In view of this, the present invention provides a method, medium and system for adjusting parameters of a dual-circulation cooling water system in a computer room, which can solve the technical problem that it is difficult to intelligently adjust various key parameters of the dual-circulation refrigeration system in the prior art.

[0009] The present invention is achieved in that:

[0010] A first aspect of the present invention provides a method for adjusting parameters of a double-circulation cooling water system in a computer room, comprising the following steps:

[0011] S10, establishing an operating equation group of the computer room double-circulation cooling water system taking into account the server computing business volume and the operating parameters of the computer room double-circulation cooling water system;

[0012] S20, fitting the operating equation group using the historically collected server computing business volume and the operating parameters of the dual-circulation cooling water system in the computer room, and obtaining the fitted operating equation group as a dual-circulation model;

[0013] S30, establishing a two-layer model according to the dual-circulation model and the real-time operation data of the system, including a multi-objective cooling capacity allocation optimization model for overall cooling capacity allocation, and a single server cooling model for precise cooling control of a single server;

[0014] S40, using the objective function of the multi-objective cooling capacity allocation optimization model as a fitness function, establishing a gray wolf hunting model, iteratively obtaining multiple non-inferior solutions, and forming a preliminary cooling capacity allocation solution group for a dual-circulation cooling water system;

[0015] S50, using the preliminary cooling capacity allocation plan group as a constraint, inputting it into the single server cooling model, iteratively optimizing to obtain the precise cooling capacity demand of each server, and forming a final cooling capacity allocation plan;

[0016] S60, calculating the target values ​​of the adjustment parameters of the double-circulation cooling water system of the computer room according to the final cooling capacity allocation plan, and outputting the target values ​​to the double-circulation cooling water system of the computer room for dynamic adjustment;

[0017] S70, executing S40-S60 in a cycle in the double-circulation refrigeration water system of the machine room.

[0018] On the basis of the above technical solution, the parameter adjustment method of a double-circulation cooling water system in a computer room of the present invention can also be improved as follows:

[0019] The operating equation group includes a refrigeration cycle equation, a cooling cycle equation, a server traffic temperature equation, a heat exchange equation and an energy consumption equation.

[0020] Furthermore, the operating parameters include chiller outlet water temperature, chiller load rate, chilled water flow rate, cooling water flow rate, cooling tower fan speed, server inlet and outlet water temperature, server computing business volume, water pump frequency, and plate heat exchanger heat exchange efficiency.

[0021] Furthermore, the objective function of the multi-objective cooling capacity allocation optimization model includes minimizing total energy consumption and maximizing cooling efficiency, and the constraints include server temperature limit, equipment operating range, and system total cooling capacity supply capacity.

[0022] Furthermore, the objective function of the single server cooling model includes minimizing server temperature fluctuations and maximizing cooling efficiency, and the constraints include server temperature limit, cooling capacity allocation upper limit, and server power limit.

[0023] Furthermore, the target values ​​of various adjustment parameters of the double-circulation refrigeration water system in the computer room include the water outlet temperature of the chiller, the chilled water flow rate, the cooling water flow rate, the cooling tower fan speed, and the water pump frequency.

[0024] Furthermore, the refrigeration cycle equation considers the heat transfer of chilled water in the double-circulation refrigeration water system of the computer room, and obtains the cooling capacity of the chilled water cycle, which is used to evaluate the refrigeration capacity and efficiency of the double-circulation refrigeration water system of the computer room;

[0025] The cooling cycle equation considers the heat discharge of cooling water in the double-circulation cooling water system of the computer room, and obtains the heat dissipation of the cooling cycle, which is used to evaluate the performance of the cooling tower of the double-circulation cooling water system of the computer room and the heat balance of the double-circulation cooling water system of the computer room;

[0026] The server business volume temperature equation takes into account the impact of server computing load on temperature, and obtains the relationship between server temperature, business volume and cooling capacity, which is used to predict and control server temperature;

[0027] The heat exchange equation considers the heat transfer process of the plate heat exchanger of the double-circulation cooling water system of the computer room, and obtains the heat exchange amount of the heat exchanger, which is used to evaluate the heat exchange efficiency of the double-circulation cooling water system of the computer room;

[0028] The energy consumption equation takes into account the energy consumption of each component of the double-circulation cooling water system of the computer room to obtain the total energy consumption of the entire double-circulation cooling water system of the computer room, which is used to evaluate the energy efficiency and optimize the operation strategy of the double-circulation cooling water system of the computer room.

[0029] Furthermore, the double-circulation refrigeration water system of the computer room includes a chilled water circulation subsystem, a cooling water circulation subsystem, a control system, a monitoring system and auxiliary equipment.

[0030] The chilled water circulation subsystem comprises:

[0031] The chiller is used to produce low-temperature chilled water and is the core refrigeration equipment of the system;

[0032] Chilled water pump, responsible for delivering chilled water and ensuring that chilled water circulates in the system;

[0033] Chilled water pipelines connect various devices and are used to transport chilled water;

[0034] Plate heat exchangers, which serve as the heat exchange interface between the chilled water loop and the server cooling loop;

[0035] The cooling water circulation subsystem comprises:

[0036] Cooling towers, which are used to dissipate heat from cooling water to the atmosphere;

[0037] Cooling water pump, responsible for delivering cooling water and ensuring that cooling water circulates in the system;

[0038] Cooling water pipelines, connecting cooling towers, chillers and other equipment, are used to transport cooling water;

[0039] The control system comprises:

[0040] Temperature sensors to measure the temperature of chilled water, cooling water, and servers;

[0041] Flow meters, used to measure the flow of chilled water and cooling water;

[0042] Frequency converters, used to adjust the speed of water pumps and cooling tower fans;

[0043] A controller, which receives sensor data, executes optimization algorithms, and outputs control instructions;

[0044] The monitoring system comprises:

[0045] Server performance monitoring software, used to monitor the server's computing workload in real time;

[0046] Energy consumption monitoring device, used to record the energy consumption data of each component of the system;

[0047] The auxiliary equipment includes:

[0048] Expansion water tank, used to absorb volume changes caused by changes in water temperature and maintain stable system pressure;

[0049] Filters are used to remove impurities from circulating water and protect system equipment;

[0050] Valve: used to adjust flow and control waterway;

[0051] The dual-circulation chilled water system in the computer room achieves efficient cooling of the computer room through the coordinated work of the chilled water cycle and the cooling water cycle; the chilled water cycle is responsible for absorbing the heat generated by the server, while the cooling water cycle is responsible for discharging this heat to the environment; through real-time monitoring and optimization of various system parameters, the energy-saving and efficient operation of the computer room can be achieved, while ensuring the normal operating temperature of the server.

[0052] Furthermore, the operating equation group in S10 is specifically expressed as follows:

[0053] Refrigeration cycle equation:

[0054] Q c =m c c p (T co -T ci )+ε c ;

[0055] In the formula, Q c is the cooling capacity of the refrigeration cycle, in kW; m c is the mass flow rate of chilled water, in kg / s; c p is the specific heat capacity of water, take 4.2 kJ / (kg·K); T co T is the chilled water outlet temperature, in °C; ci is the chilled water inlet temperature, in °C; ε c is the error term.

[0056] Cooling cycle equation: Q h =m h c p (T hi -T ho )+ε h ;

[0057] In the formula, Q h is the heat dissipation of the cooling cycle, in kW; m h is the cooling water mass flow rate, in kg / s; T hi T is the cooling water inlet temperature, in °C; ho is the cooling water outlet temperature, in °C; ε h is the error term.

[0058] Server traffic temperature equation:

[0059] T s =T s0 +k1L+k2Q c +ε s ;

[0060] Where, T s is the server temperature, in °C; T s0 is the initial temperature of the server, in °C; L is the server computing workload, in FLOPS; k1, k2 are unknown coefficients; ε s is the error term.

[0061] Heat exchange equation:

[0062] Q e =UAΔT lm +ε e ;

[0063] In the formula, Q eis the heat transfer capacity of the heat exchanger, in kW; U is the total heat transfer coefficient, in kW / (m2·K); A is the heat transfer area, in m 2 ; ΔT lm is the logarithmic mean temperature difference, in °C; ε e is the error term.

[0064] Energy consumption equation:

[0065] E=E c +E p +E t +ε E ;

[0066] Where, E is the total energy consumption, in kWh; E c is the energy consumption of the chiller, in kWh; E p is the energy consumption of the water pump, in kWh; E t is the energy consumption of the cooling tower, in kWh; ε E is the error term.

[0067] Parameter acquisition method:

[0068] m c ,m h : Obtained through real-time measurement by flow meter;

[0069] T co ,T ci ,T hi ,T ho : Obtained through real-time measurement by temperature sensor;

[0070] L: Obtained in real time through server performance monitoring software;

[0071] U: Calculated by plate heat exchanger parameters, the calculation formula is: Where h1, h2 are the convective heat transfer coefficients on the cold and hot sides, in kW / (m^2·K); δ is the thickness of the heat transfer plate, in m; λ is the thermal conductivity of the heat transfer plate, in kW / (m·K);

[0072] ΔT lm The calculation formula is:

[0073] The fitting method in S20 adopts the least squares method, and the specific steps are as follows:

[0074] Construct the error sum of squares function: where y i is the actual observed value, is the model prediction value;

[0075] Find the parameter value that minimizes SSE: where β j is a parameter to be determined;

[0076] Solve the linear system to obtain parameter estimates.

[0077] The multi-objective cooling capacity allocation optimization model in S30 is specifically expressed as follows:

[0078] Objective function:

[0079]

[0080] Constraints:

[0081] T s,min ≤T s ≤T s,max ;

[0082] Q c,min ≤Q c ≤Q c,max ;

[0083]

[0084] In the formula, E i is the energy consumption of the ith device, in kWh; Q ci is the cooling capacity of the i-th device, in kW; T s,min ,T s,max The lowest and highest temperatures allowed by the server, in °C; Q c,min ,Q c,max are the minimum and maximum cooling capacity of the chiller, in kW; Q total is the total cooling capacity of the system in kW.

[0085] The single server cooling model is specifically expressed as follows:

[0086] Objective function:

[0087]

[0088] Constraints:

[0089] T s,min ≤T s ≤T s,max ;

[0090] Q c ≤Q c,allocated ;

[0091] P≤P max ;

[0092] Where, T s (t) is the server temperature at time t, in °C; T s,targetis the target temperature, in °C; T is the time step; Q c,allocated is the cooling capacity allocated to the server, in kW; P is the server power, in kW; P max The maximum power of the server in kW.

[0093] The gray wolf hunting model in S40 is specifically expressed as follows:

[0094] Position update formula:

[0095]

[0096] In the formula, is the wolf pack position vector at time t; is the prey position vector at time t; is the coefficient vector; is the convergence factor, which decreases linearly with the number of iterations and ranges from [0,2]; is a random vector between [0,1].

[0097] The iterative optimization in S50 adopts the gradient descent method, and the specific steps are as follows:

[0098] Step 1. Initialize parameter θ;

[0099] Step 2. Calculate the gradient of the objective function J(θ):

[0100] Step 3. Update parameters: Where α is the learning rate;

[0101] Step 4. Repeat steps 2-3 until convergence.

[0102] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the above-mentioned method for adjusting parameters of a double-circulation chilled water system in a computer room.

[0103] A third aspect of the present invention provides a parameter adjustment system for a dual-circulation chilled water system in a computer room, which includes the above-mentioned computer-readable storage medium.

[0104] Compared with the prior art, the method, medium and system for adjusting parameters of a double-circulation cooling water system in a computer room provided by the present invention have the following beneficial effects:

[0105] 1) Accurately match the real-time cooling needs of the server. This method establishes a complete set of operating equations including the server business volume temperature model, which can accurately predict the server temperature changes and dynamically adjust the parameters of the cooling system accordingly to ensure that the server temperature is always maintained within the ideal range.

[0106] 2) Balance between system energy consumption and refrigeration performance. This method establishes a multi-objective optimization model, which simultaneously considers the two objectives of minimizing total energy consumption and maximizing refrigeration efficiency. Through the iterative optimization of the gray wolf hunting algorithm, the optimal solution that meets the dual requirements of energy efficiency and refrigeration performance can be found.

[0107] 3) Global optimization of the system operation status is achieved. This method couples the optimization of cooling capacity allocation and single server cooling control, fully considers the overall performance of the system, and fine-tunes the cooling capacity demand of each server through the gradient descent method to achieve the global optimum.

[0108] 4) Good adaptability and robustness. The dual-loop model established based on historical data can continuously track changes in the system's operating status and dynamically adjust the optimization strategy to ensure that the best performance can be maintained under various working conditions.

[0109] In summary, the adaptive adjustment method of the computer room dual-circulation refrigeration system based on multi-objective optimization proposed in the present invention can accurately respond to changes in server load and achieve a high balance between energy consumption and refrigeration performance, providing a more intelligent and efficient refrigeration solution for the data center, and solving the technical problem that the existing technology is difficult to intelligently adjust the key parameters of the dual-circulation refrigeration system. BRIEF DESCRIPTION OF THE DRAWINGS

[0110] Figure 1 A flow chart of the method provided by the present invention;

[0111] Figure 2 This is a diagram showing the changes in server temperature before and after optimization;

[0112] Figure 3 This is a comparison chart of the total energy consumption of the system before and after optimization;

[0113] Figure 4 This is a diagram showing the changes in the system's cooling performance before and after optimization. DETAILED DESCRIPTION

[0114] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0115] like Figure 1 As shown, it is a flow chart of a method for adjusting parameters of a double-circulation cooling water system in a computer room provided by the present invention. The method comprises the following steps:

[0116] S10, establishing an operating equation group of the computer room double-circulation cooling water system taking into account the server computing business volume and the operating parameters of the computer room double-circulation cooling water system;

[0117] S20, fitting the operating equation group using the historically collected server computing business volume and the operating parameters of the dual-circulation cooling water system in the computer room, and obtaining the fitted operating equation group as a dual-circulation model;

[0118] S30, establishing a two-layer model according to the dual-circulation model and the real-time operation data of the system, including a multi-objective cooling capacity allocation optimization model for overall cooling capacity allocation, and a single server cooling model for precise cooling control of a single server;

[0119] S40, using the objective function of the multi-objective cooling capacity allocation optimization model as the fitness function, establishing a gray wolf hunting model, iteratively obtaining multiple non-inferior solutions, and forming a preliminary cooling capacity allocation solution group for the dual-circulation cooling water system;

[0120] S50, using the preliminary cooling capacity allocation plan group as a constraint, inputting it into the single server cooling model, iteratively optimizing to obtain the precise cooling capacity demand of each server, and forming a final cooling capacity allocation plan;

[0121] S60, according to the final cooling capacity allocation plan, the target values ​​of various adjustment parameters of the double-circulation cooling water system of the computer room are output to the double-circulation cooling water system of the computer room for dynamic adjustment;

[0122] S70. Execute S40-S60 in a double-circulation cooling water system in the computer room.

[0123] The specific implementation methods of the above steps are described in detail below:

[0124] The specific implementation of step S10 is: first, establish an operating equation group of the computer room dual-circulation cooling water system that takes into account the server computing business volume and the operating parameters of the computer room dual-circulation cooling water system. The operating equation group includes five main equations, namely, the refrigeration cycle equation, the cooling cycle equation, the server business volume temperature equation, the heat exchange equation, and the energy consumption equation. These five equations cover various key characteristics of the computer room dual-circulation cooling water system and can comprehensively describe the operating status of the system.

[0125] Refrigeration cycle equation: Q c =m c c p (T co -T ci )+ε c ; In the refrigeration cycle equation, Q c Indicates the cooling capacity of the refrigeration cycle, in kW; m c Indicates the mass flow rate of chilled water, in kg / s; c p represents the specific heat capacity of water, which is 4.2 kJ / (kg·K); T co Indicates the chilled water outlet temperature in °C; T ciIndicates the chilled water inlet temperature, in °C; ε c Represents the error term. This equation describes the relationship between the cooling capacity of the refrigeration cycle and the chilled water flow and temperature difference.

[0126] Cooling cycle equation: Q h =m h c p (T hi -T ho )+ε h ; In the cooling cycle equation, Q h Indicates the heat dissipation of the cooling cycle, in kW; m h Indicates the cooling water mass flow rate in kg / s; T hi Indicates the cooling water inlet temperature, in °C; T ho Indicates the cooling water outlet temperature, in °C; ε h Represents the error term. This equation describes the relationship between the heat dissipation of the cooling cycle and the cooling water flow and temperature difference.

[0127] Server traffic temperature equation: T s =T s0 +k1L+k2Q c +ε s ; In the server traffic temperature equation, T s Indicates the server temperature in °C; T s0 represents the initial temperature of the server, in °C; L represents the computing capacity of the server, in FLOPS; k1, k2 are unknown coefficients; ε s represents the error term. This equation describes the impact of server temperature on server computing workload and cooling capacity.

[0128] Heat exchange equation: Q e =UAΔT lm +ε e ; In the heat exchange equation, Q e It represents the heat transfer capacity of the heat exchanger, in kW; U represents the total heat transfer coefficient, in kW / (m 2 ·K); A represents the heat transfer area, the unit is m 2 ; ΔT lm Represents the logarithmic mean temperature difference, in °C; ε e represents the error term. This equation describes the heat transfer performance of the heat exchanger.

[0129] Energy consumption equation: E = E c +E p +E t +ε E ; In the energy consumption equation, E represents the total energy consumption in kWh; E cIndicates the energy consumption of the chiller, in kWh; E p Indicates the energy consumption of the water pump, in kWh; E t Indicates the energy consumption of the cooling tower, in kWh; ε E Represents the error term. This equation describes the total energy consumption of the double-cycle chilled water system in the computer room.

[0130] The operating parameters involved in these equations include chiller outlet water temperature, chiller load rate, chilled water flow, cooling water flow, cooling tower fan speed, server inlet and outlet water temperature, server computing business volume, water pump frequency, plate heat exchanger heat exchange efficiency, etc. These parameters can be obtained through real-time monitoring or calculation. Specifically, the parameter acquisition method is:

[0131] m c ,m h : Obtained through real-time measurement by flow meter;

[0132] T co ,T ci ,T hi ,T ho : Obtained through real-time measurement by temperature sensor;

[0133] L: Obtained in real time through server performance monitoring software;

[0134] U: Calculated by plate heat exchanger parameters, the calculation formula is: Where h1, h2 are the convective heat transfer coefficients on the cold and hot sides, in kW / (m^2·K); δ is the thickness of the heat transfer plate, in m; λ is the thermal conductivity of the heat transfer plate, in kW / (m·K);

[0135] ΔT lm The calculation formula is:

[0136] The specific implementation method of step S20 is: fitting the above-mentioned operating equation group using the historically collected server computing business volume and the operating parameters of the dual-circulation cooling water system in the computer room to obtain the fitted operating equation group, which is recorded as the dual-circulation model. The fitting method used here is the least squares method. The specific steps are as follows:

[0137] First, construct the error sum of squares function where y i is the actual observed value, is the model's predicted value.

[0138] Then, find the parameter value that minimizes SSE, that is, where β j To be determined parameter.

[0139] Finally, the linear equations are solved to obtain parameter estimates.

[0140] Through this fitting method, a double-circulation model that can accurately describe the operating characteristics of the double-circulation cooling water system in the computer room can be established based on historical data.

[0141] The specific implementation method of step S30 is: first, based on the above-mentioned dual-circulation model and the real-time operation data of the system, a double-layer model is established, wherein the upper layer model is a multi-objective cooling capacity allocation optimization model for overall cooling capacity allocation, and the lower layer model is a single server cooling model for precise cooling control of a single server.

[0142] The objective functions of the multi-objective cooling capacity allocation optimization model include two aspects: minimizing the total energy consumption and maximize cooling efficiency Among them, E i represents the energy consumption of the i-th device, in kWh; Q ci Represents the cooling capacity of the i-th device, in kW.

[0143] The constraints of the model include: server temperature limit T s,min ≤T s ≤T s,max , Equipment operating range Q c,min ≤Q c ≤Q c,max , and the total cooling capacity of the system Among them, T s,min ,T s,max Respectively represent the lowest and highest temperatures allowed by the server, in °C; Q c,min ,Q c,max Respectively represent the minimum and maximum cooling capacity of the chiller, in kW; Q total Indicates the total cooling capacity of the system in kW.

[0144] The objective function of the single server cooling model also includes two aspects: minimizing server temperature fluctuations and maximize cooling efficiency Among them, T s (t) represents the server temperature at time t, in °C; T s,target represents the target temperature in °C; T represents the time step; Q c It indicates the cooling capacity allocated to the server in kW; E indicates the corresponding energy consumption in kWh.

[0145] The constraints of the model include: server temperature limit T s,min ≤T s ≤T s,max , cooling capacity allocation upper limit Q c ≤Qc,allocated , and the server power limit P≤P max Among them, Q c,allocated Indicates the cooling capacity allocated to the server, in kW; P indicates the server power, in kW; P max Indicates the maximum power of the server in kW.

[0146] The double-layer model established in this way can achieve optimal distribution of the overall cooling capacity of the computer room, while precisely controlling the temperature of individual servers to maximize the energy efficiency and cooling performance requirements of the system.

[0147] The specific implementation method of step S40 is: using the objective function of the multi-objective cooling allocation optimization model as the fitness function, establishing a gray wolf hunting algorithm model, obtaining multiple non-inferior solutions through iterative optimization, and forming a preliminary cooling allocation solution group.

[0148] The gray wolf hunting algorithm is a heuristic optimization algorithm that simulates the behavioral characteristics of gray wolves during hunting. The position update formula in the algorithm is as follows:

[0149]

[0150] in, represents the wolf pack position vector at time t; represents the prey position vector at time t; is the coefficient vector; is the convergence factor, which decreases linearly with the number of iterations and ranges from [0,2]; is a random vector between [0,1].

[0151] Through this iterative optimization method, multiple non-inferior solutions that meet multi-objective optimization can be obtained, that is, preliminary cooling capacity allocation scheme groups. These scheme groups provide a good starting point for further optimization in the next step.

[0152] The specific implementation method of step S50 is: using the preliminary cooling capacity allocation plan group obtained in the previous step as a constraint, inputting it into the single server cooling model, using the gradient descent method for iterative optimization, obtaining the precise cooling capacity demand of each server, and forming the final cooling capacity allocation plan.

[0153] Gradient descent method is a commonly used iterative optimization algorithm, and its specific steps are as follows:

[0154] 1. Initialize the parameter θ;

[0155] 2. Calculate the gradient of the objective function J(θ):

[0156] 3. Update parameters: Where α is the learning rate;

[0157] 4. Repeat steps 2-3 until convergence.

[0158] Through this iterative optimization method, the cooling demand of each server can be further fine-tuned on the premise of satisfying the preliminary cooling allocation plan to minimize server temperature fluctuations and maximize cooling efficiency. The cooling allocation plan finally obtained is the input required in step S60.

[0159] The specific implementation of step S60 is: according to the final cooling capacity allocation plan obtained in step S50, the target values ​​of various adjustment parameters of the system are calculated, including the outlet water temperature of the chiller, the chilled water flow rate, the cooling water flow rate, the cooling tower fan speed, the water pump frequency, etc. These target values ​​are then output to the double-circulation chilled water system of the computer room to achieve dynamic adjustment.

[0160] The calculation formulas for these adjustment parameters are as follows:

[0161] Chiller outlet water temperature: T co =f(Q c1 ,Q c2 ,…,Q cn );

[0162] Chilled water flow:m c =f(Q c1 ,Q c2 ,…,Q cn );

[0163] Cooling water flow:m h =f(Q h1 ,Q h2 ,…,Q hn );

[0164] Cooling tower fan speed: ω t =f(T hi ,T ho );

[0165] Pump frequency: f p =f(m c ,m h );

[0166] Among them, Q ci ,Q hi Respectively represent the cooling capacity and heat dissipation of the i-th device, in kW; T hi ,T ho Respectively represent the inlet and outlet temperatures of cooling water, in °C; ω t Indicates the cooling tower fan speed, f p Indicates the pump frequency.

[0167] By precisely adjusting these key parameters, it is possible to ensure that the computer room's dual-circulation chilled water system operates according to the optimal plan, thereby achieving the goal of minimizing energy consumption and maximizing cooling performance.

[0168] Step S70 is to execute steps S40-S60 in a loop throughout the whole process, continuously optimizing and adjusting the cooling capacity distribution plan to keep the system in the best operating state. Through this dynamic optimization method, it can fully adapt to the changing server computing business volume and continuously improve the energy efficiency and cooling performance of the computer room dual-circulation cooling water system.

[0169] In general, this parameter adjustment method based on the operating equations of the computer room dual-circulation cooling water system, multi-objective optimization model and iterative algorithm can comprehensively consider various performance indicators of the system, achieve the goals of minimizing energy consumption and maximizing cooling efficiency, and is an effective computer room cooling system optimization solution. This method makes full use of existing monitoring data and modeling technology, and provides strong support for the intelligent management of the computer room cooling system through multi-level optimization strategies.

[0170] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the above-mentioned method for adjusting parameters of a double-circulation chilled water system in a computer room.

[0171] A third aspect of the present invention provides a parameter adjustment system for a dual-circulation chilled water system in a computer room, which includes the above-mentioned computer-readable storage medium.

[0172] Specifically, the principle of the present invention is:

[0173] 1. Establish a global operation model of the dual-cycle refrigeration system in the computer room. The model includes mathematical descriptions of key processes such as refrigeration cycle, cooling cycle, server temperature changes, heat exchange, etc., covering all components of the system. By fitting historical data, a dual-cycle model that can accurately describe the operating characteristics of the system can be obtained. This lays the foundation for subsequent optimization analysis.

[0174] 2. Adopt a multi-objective optimization cooling capacity allocation strategy. Traditional single-objective optimization either focuses on minimizing energy consumption or maximizing cooling performance, and it is difficult to take both into account. The present invention adopts a multi-objective optimization approach, takes minimizing total energy consumption and maximizing cooling efficiency as objective functions, and uses the gray wolf hunting algorithm to find the optimal solution that meets these two objectives. This ensures that the cooling demand is met and minimizes system energy consumption.

[0175] 3. Introduce a two-layer optimization mechanism. The optimization strategy of the present invention is divided into two layers: the upper layer is based on the cooling capacity allocation optimization of the entire system, and the lower layer is the fine temperature control for a single server. The multi-objective optimization of the upper layer determines the overall cooling capacity demand of each device, and the single-server model of the lower layer further optimizes the cooling capacity allocation of each server, achieving global and local collaborative optimization. This two-layer structure can fully reflect the optimization of the overall performance of the system.

[0176] 4. Adopt adaptive optimization algorithms. The optimization algorithms involved in the method of the present invention, such as the gray wolf hunting algorithm and the gradient descent method, have good adaptability. They can automatically adjust the optimization strategy according to the changes in the system operation state to ensure that the best performance can be maintained under various working conditions. At the same time, these algorithms also provide feasibility for real-time optimization and can quickly respond to dynamic changes in server load.

[0177] In order to better understand and implement the present invention, a specific embodiment of the present invention is provided below: A data center of a large Internet company is using a double-circulation chilled water system in a computer room. The data center covers an area of ​​20,000 square meters, has a total of 8 computer rooms, and is equipped with 1,200 servers. The refrigeration system uses 2 100RT (cooling capacity of about 350kW) centrifugal chillers, equipped with 2 300HP cooling water pumps and 4 150HP freezing water pumps. At the same time, there are also 4 100kW cooling towers.

[0178] Recently, with the rapid growth of business, the computing load of servers has fluctuated significantly, and the operation of the refrigeration system has also faced many challenges: first, it is difficult to adjust the cooling capacity in time to meet the real-time needs of the server, resulting in excessive temperatures on some servers; second, manual adjustment of refrigeration parameters is too empirical and it is difficult to strike a balance between energy consumption and refrigeration performance; third, there is a lack of global optimization of the overall performance of the system, which affects the overall efficiency of the refrigeration system.

[0179] In view of the above problems, the enterprise decided to adopt the adaptive optimization method of the double-circulation cooling water system in the computer room proposed in this invention to upgrade the existing refrigeration system. The following is the specific implementation process:

[0180] 1. Establish the operation model of the double-circulation cooling water system in the computer room

[0181] Firstly, based on the on-site monitoring data, the operating equations of the computer room's double-circulation chilled water system were established.

[0182] These include:

[0183] Refrigeration cycle equation:

[0184] Q c =350×(T co -T ci );

[0185] In the formula, Q c is the cooling capacity of the refrigeration cycle, in kW; T co T is the chilled water outlet temperature, in °C; ci is the chilled water inlet temperature, in °C.

[0186] Cooling cycle equation:

[0187] Q h =350×(T hi -T ho );

[0188] In the formula, Q h is the heat dissipation of the cooling cycle, in kW; T hi T is the cooling water inlet temperature, in °C; ho is the cooling water outlet temperature, in °C.

[0189] Server traffic temperature equation:

[0190] T s =25+0.0001L+0.2Q c ;

[0191] Where, T s is the server temperature, in °C; L is the server computing workload, in FLOPS.

[0192] Heat exchange equation:

[0193]

[0194] In the formula, Q e is the heat exchanger heat capacity, unit is kW.

[0195] Energy consumption equation:

[0196] E = 100 × (P c +P p +P t );

[0197] Where, E is the total energy consumption, in kWh; P c ,P p ,P t They are the power of chiller, water pump and cooling tower respectively, in kW.

[0198] By fitting historical data, the parameter values ​​of the above equation were obtained, and a more accurate dual-circulation operation model was established.

[0199] 2. Establish a two-layer optimization model

[0200] Based on the above double-circulation operation model, the method of the present invention establishes a two-layer optimization model:

[0201] The first layer is a multi-objective cooling capacity allocation optimization model, and the objective functions include:

[0202] Minimize total energy consumption

[0203] Maximize cooling efficiency

[0204] Among them, E i is the energy consumption of the i-th server, in kWh; Q ci is the cooling capacity of the i-th server, in kW.

[0205] Constraints include:

[0206] Server temperature limit: 20 ≤ T s ≤30;

[0207] Equipment operating range: 320≤Q c ≤380,280≤Q h ≤320;

[0208] Total cooling capacity of the system:

[0209] The second layer is a single server precise cooling model, and the objective functions include:

[0210] Minimize temperature fluctuations

[0211] Maximize cooling efficiency

[0212] Constraints include:

[0213] Server temperature limit: 20 ≤ T s ≤30;

[0214] Cooling capacity allocation upper limit: Q c ≤Q c,allocated ;

[0215] Server power limit: P≤80;

[0216] Such a two-layer optimization model can achieve overall optimization of the entire system and fine control of a single server.

[0217] 3. Use the gray wolf hunting algorithm to optimize cooling distribution

[0218] Based on the above multi-objective optimization model, the present invention uses the gray wolf hunting algorithm to optimize the cooling allocation scheme. The specific steps are as follows:

[0219] 1) Initialize the wolf pack position vector and the prey position vector

[0220] 2) Calculate the fitness value of each wolf, that is, the objective function value of the multi-objective optimization model.

[0221] 3) Update the positions of alpha wolf, beta wolf and delta wolf.

[0222] 4) Update the formula based on the position in Updates the location of other wolves. is a linearly decreasing convergence factor.

[0223] 5) Determine whether the convergence condition is met. If not, return to step 2) and continue iterating.

[0224] Through the above iterative optimization, multiple non-inferior solutions that meet the requirements of minimizing energy consumption and maximizing cooling efficiency can be obtained, that is, a preliminary group of cooling capacity allocation solutions.

[0225] 4. Use gradient descent method to optimize single server cooling

[0226] Taking the preliminary cooling capacity allocation scheme obtained in the previous step as a constraint, the present invention uses the gradient descent method to further optimize the cooling capacity demand of each server. The specific steps are as follows:

[0227] 1) Initialize the cooling capacity allocation Q of each server c .

[0228] 2) Calculate the objective function value and its gradient of the single server cooling model.

[0229] 3) According to the gradient descent formula Update the cooling capacity allocation, where α is the learning rate.

[0230] 4) Determine whether the convergence condition is met. If not, return to step 2) and continue iterating.

[0231] Through this iterative optimization, the precise cooling demand of each server can be obtained to form the final cooling allocation plan.

[0232] 5. Dynamically adjust refrigeration system parameters

[0233] According to the final cooling capacity allocation scheme obtained by the above optimization, the present invention calculates the target values ​​of various adjustment parameters of the system, including:

[0234] Chiller outlet water temperature T co =26℃;

[0235] Chilled water flow mc =180kg / s;

[0236] Cooling water flow m h =210kg / s;

[0237] Cooling tower fan speedω t =80%;

[0238] Pump frequency f p =50Hz;

[0239] These parameter values ​​are output to the refrigeration system through the automatic control system to achieve dynamic adjustment.

[0240] The effect of the entire optimization process is as follows:

[0241] 1) The server temperature control effect is significantly improved. By accurately matching the server cooling demand, the server temperature is always maintained within the ideal range of 25±2℃, which is far better than the large fluctuations during manual adjustment.

[0242] 2) System energy consumption is significantly reduced. The total energy consumption after optimization is reduced from the original 1450kWh to 1250kWh, a decrease of 13.8%, reflecting a significant improvement in energy efficiency.

[0243] 3) Refrigeration performance is guaranteed. After optimization, the cooling efficiency of the system is increased from the original 2.8 to 3.2, providing a strong guarantee for meeting high-load business.

[0244] The advantages of the method of the present invention are further demonstrated by several figures below: Figure 1 The change of server temperature before and after optimization is shown. It can be seen that the server temperature fluctuated greatly during manual adjustment, exceeding the allowable temperature range, while the temperature was always maintained within the ideal range after the method of the present invention was adopted, and the response to server load changes was more rapid and effective. Figure 2 The total energy consumption of the system before and after optimization is compared. It can be seen that after adopting the method of the present invention, the total energy consumption of the system is significantly reduced, reflecting good energy efficiency performance. Figure 3 It reflects the change in the cooling performance of the system before and after optimization. It can be seen that the cooling efficiency of the optimized system has been significantly improved, providing better cooling guarantee for high-load business.

[0245] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for adjusting parameters of a double-circulation cooling water system in a computer room, characterized in that: The following steps are involved: S10, establishing an operating equation group of the computer room double-circulation cooling water system taking into account the server computing business volume and the operating parameters of the computer room double-circulation cooling water system; S20, fitting the operating equation group using the historically collected server computing business volume and the operating parameters of the dual-circulation cooling water system in the computer room, and obtaining the fitted operating equation group as a dual-circulation model; S30, establishing a two-layer model according to the dual-circulation model and the real-time operation data of the system, including a multi-objective cooling capacity allocation optimization model for overall cooling capacity allocation, and a single server cooling model for precise cooling control of a single server; S40, using the objective function of the multi-objective cooling capacity allocation optimization model as a fitness function, establishing a gray wolf hunting model, iteratively obtaining multiple non-inferior solutions, and forming a preliminary cooling capacity allocation solution group for a dual-circulation cooling water system; S50, using the preliminary cooling capacity allocation plan group as a constraint, inputting it into the single server cooling model, iteratively optimizing to obtain the precise cooling capacity demand of each server, and forming a final cooling capacity allocation plan; S60, calculating the target values ​​of the adjustment parameters of the double-circulation cooling water system of the computer room according to the final cooling capacity allocation plan, and outputting the target values ​​to the double-circulation cooling water system of the computer room for dynamic adjustment; S70, executing S40-S60 in a cycle in the double-circulation refrigeration water system of the machine room.

2. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 1, characterized in that: The operating equation group includes a refrigeration cycle equation, a cooling cycle equation, a server traffic temperature equation, a heat exchange equation, and an energy consumption equation.

3. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 2, characterized in that: The operating parameters include chiller outlet water temperature, chiller load rate, chilled water flow rate, cooling water flow rate, cooling tower fan speed, server inlet and outlet water temperature, server computing business volume, water pump frequency, and plate heat exchanger heat exchange efficiency.

4. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 3, characterized in that: The objective function of the multi-objective cooling capacity allocation optimization model includes minimizing total energy consumption and maximizing cooling efficiency. The constraints include server temperature limit, equipment operating range, and system total cooling capacity supply capacity.

5. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 4, characterized in that: The objective function of the single server cooling model includes minimizing server temperature fluctuation and maximizing cooling efficiency, and the constraints include server temperature limit, cooling capacity allocation upper limit, and server power limit.

6. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 5, characterized in that: The target values ​​of various adjustment parameters of the double-circulation refrigeration water system in the computer room include the water outlet temperature of the chiller, the chilled water flow rate, the cooling water flow rate, the cooling tower fan speed, and the water pump frequency.

7. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 6, characterized in that: The refrigeration cycle equation considers the heat transfer of chilled water in the double-circulation refrigeration water system of the computer room, and obtains the cooling capacity of the chilled water cycle, which is used to evaluate the refrigeration capacity and efficiency of the double-circulation refrigeration water system of the computer room; The cooling cycle equation considers the heat discharge of cooling water in the double-circulation cooling water system of the computer room, and obtains the heat dissipation of the cooling cycle, which is used to evaluate the performance of the cooling tower of the double-circulation cooling water system of the computer room and the heat balance of the double-circulation cooling water system of the computer room; The server business volume temperature equation takes into account the impact of server computing load on temperature, and obtains the relationship between server temperature, business volume and cooling capacity, which is used to predict and control server temperature; The heat exchange equation considers the heat transfer process of the plate heat exchanger of the double-circulation cooling water system of the computer room, and obtains the heat exchange amount of the heat exchanger, which is used to evaluate the heat exchange efficiency of the double-circulation cooling water system of the computer room; The energy consumption equation takes into account the energy consumption of each component of the double-circulation cooling water system of the computer room to obtain the total energy consumption of the entire double-circulation cooling water system of the computer room, which is used to evaluate the energy efficiency and optimize the operation strategy of the double-circulation cooling water system of the computer room.

8. A method for adjusting parameters of a double-circulation cooling water system in a computer room according to claim 7, characterized in that: The double-circulation refrigeration water system of the computer room includes a chilled water circulation subsystem, a cooling water circulation subsystem, a control system, a monitoring system and auxiliary equipment.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the method for adjusting parameters of a dual-circulation chilled water system in a computer room according to any one of claims 1 to 8.

10. A parameter adjustment system for a double-circulation cooling water system in a computer room, characterized in that: A computer-readable storage medium comprising the computer-readable storage medium of claim 9.