Data center machine room airflow organization optimization method based on proxy model

Through the multi-objective optimization method based on the agent model, the problem of single airflow organization optimization goals in the data center cooling system is solved, efficient multi-objective optimization is achieved, computing resource requirements are reduced, and the optimal airflow organization distribution is found.

CN118940665APending Publication Date: 2024-11-12CHINA YANGTZE POWER +1
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202410996689.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

In the prior art, it is difficult to consider multiple airflow organization distribution goals at the same time, resulting in low optimization efficiency, wasted computing resources and single targets.

Method used

Using the agent model-based method, the mapping relationship between air supply parameters and airflow organization evaluation indicators is constructed, and the target air supply parameters are determined to achieve multi-objective optimization of airflow organization.

Benefits of technology

It effectively reduces the workload of numerical simulation, reduces the computing resource requirements, and can consider multiple design variables and target variables at the same time to find the optimal distribution state of the airflow tissue, improving the optimization efficiency of the airflow tissue.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940665A_ABST
    Figure CN118940665A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of computers, and particularly discloses a data center machine room airflow organization optimization method based on an agent model, and the method comprises the steps: carrying out the evaluation of machine room airflow organization simulation data based on each airflow organization evaluation index and the machine room airflow organization simulation data corresponding to each sample point in a design space, determining airflow organization evaluation data corresponding to each sample point, wherein the design space is determined based on a value range corresponding to each air supply parameter; on the basis of the airflow organization evaluation data corresponding to the sample points, an agent model is constructed, and the agent model is used for representing the mapping relation between the air supply parameters and airflow organization evaluation indexes; and target air supply parameters are determined through multi-target optimization based on the airflow organization evaluation indexes and the agent model. According to the method, the technical means of constructing the agent model and the technical means of multi-objective optimization are combined, so that multiple objectives of efficiently optimizing the airflow organization of the machine room can be realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the field of computer technology, and more specifically, to a method for optimizing airflow organization in a data center computer room based on an agent model. Background Art

[0002] For a data center with a determined cooling source, the factor that most directly affects energy consumption is the distribution of airflow in the room.

[0003] In the prior art, the optimization design research of data center cooling system generally uses computational fluid dynamics (CFD) simulation software for offline simulation, but the numerical simulation has large amount of calculation and high time cost. In addition, the goal of optimizing the airflow organization in the computer room is single, and it is difficult to find the optimal airflow organization distribution state. How to efficiently optimize the multiple goals of the airflow organization in the computer room is a technical problem that needs to be solved in this field. Summary of the invention

[0004] In view of the defects of the prior art, the purpose of this application is to achieve multiple goals of efficiently optimizing the airflow organization of the computer room.

[0005] To achieve the above objectives, in a first aspect, the present application provides a method for optimizing airflow organization in a data center computer room based on an agent model, comprising:

[0006] Based on each airflow organization evaluation index and the computer room airflow organization simulation data corresponding to each sample point in the design space, the computer room airflow organization simulation data is evaluated to determine the airflow organization evaluation data corresponding to each sample point. The design space is determined based on the value range corresponding to each air supply parameter.

[0007] Based on the airflow organization evaluation data corresponding to each sample point, a proxy model is constructed, and the proxy model is used to characterize the mapping relationship between the air supply parameters and the airflow organization evaluation index;

[0008] Based on various airflow organization evaluation indicators and proxy models, the target air supply parameters are determined through multi-objective optimization.

[0009] In a possible implementation, based on various airflow organization evaluation indicators and proxy models, target air supply parameters are determined through multi-objective optimization, including:

[0010] Based on various airflow organization evaluation indicators and agent models, an objective function is constructed, with air supply parameters as decision variables of the objective function and airflow organization evaluation indicators as optimization targets of the objective function.

[0011] Based on the objective function, the multi-objective optimization of the airflow organization in the computer room is carried out through the second generation non-dominated sorting genetic algorithm to determine the target air supply parameters.

[0012] In a possible implementation, before evaluating the airflow organization simulation data of the computer room, the following steps are further included:

[0013] For each sample point, the airflow organization of the computer room is simulated by using the computational fluid dynamics (CFD) simulation tool to obtain the simulation data of the airflow organization of the computer room corresponding to each sample point.

[0014] In a possible implementation, for each sample point, the airflow organization of the computer room is simulated by a computational fluid dynamics (CFD) simulation tool to obtain the simulation data of the airflow organization of the computer room corresponding to each sample point, including performing the following operations for each sample point:

[0015] Obtain a computer room simulation model, which is used to characterize computer room cabinets and ventilation methods;

[0016] Based on the sample points, set the operating parameters of the computer room simulation model;

[0017] Based on the computer room simulation model after setting the working condition parameters, the computational fluid dynamics (CFD) simulation tool is used to simulate the fluid movement and obtain the computer room airflow organization simulation data corresponding to the sample points.

[0018] In a possible implementation, before simulating the airflow organization of the computer room by using a computational fluid dynamics (CFD) simulation tool, the following is also included:

[0019] Sampling is performed in the design space using the Latin hypercube sampling method to obtain multiple sample points.

[0020] In a possible implementation, the air supply parameter is air supply temperature, air supply speed or air supply angle.

[0021] In a possible implementation, the airflow organization evaluation index is a heating index SHI, a heat mixing index HMI, or a rack inlet and exhaust air average temperature difference index β;

[0022] The heating index SHI is used to characterize the degree of mixing of the supply air with the hot air before entering the cold aisle;

[0023] The thermal mixing index HMI is used to characterize the worst airflow mixing condition of the rack in the cabinet;

[0024] The average temperature difference index β between the rack inlet and exhaust air is used to evaluate the possibility of the existence of local hot spots.

[0025] In a second aspect, the present application provides a data center computer room airflow organization optimization device based on an agent model, comprising:

[0026] A simulation data evaluation module is used to evaluate the airflow organization simulation data of the computer room based on each airflow organization evaluation index and the airflow organization simulation data of the computer room corresponding to each sample point in the design space, and determine the airflow organization evaluation data corresponding to each sample point. The design space is determined based on the value range corresponding to each air supply parameter;

[0027] An agent model building module is used to build an agent model based on the airflow organization evaluation data corresponding to each sample point. The agent model is used to characterize the mapping relationship between the air supply parameters and the airflow organization evaluation index;

[0028] The multi-objective optimization module is used to determine the target air supply parameters through multi-objective optimization based on various airflow organization evaluation indicators and proxy models.

[0029] In a third aspect, the present application provides an electronic device comprising: at least one memory for storing programs; and at least one processor for executing the programs stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method described in the first aspect or any possible implementation of the first aspect.

[0030] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method described in the first aspect or any possible implementation of the first aspect.

[0031] In general, the above technical solutions conceived by this application have the following beneficial effects compared with the prior art:

[0032] By constructing a proxy model (a simplified mathematical model) to approximate the real physical process, the airflow organization and related evaluation indicators under different operating parameters can be quickly solved (compared to simulating the airflow organization of the computer room through CFD simulation tools), which can effectively reduce the workload of numerical simulation and reduce the computing resources required in the optimization process. Based on various airflow organization evaluation indicators and proxy models, multi-objective optimization can be performed, which can simultaneously consider multiple design variables (air supply parameters as design variables) and target variables (airflow organization evaluation indicators as target variables) of the computer room airflow organization, avoid the single goal of the computer room airflow organization optimization, and easily find the optimal airflow organization distribution state. Therefore, by combining the technical means of constructing a proxy model with the technical means of multi-objective optimization, it is possible to achieve multiple goals of efficiently optimizing the airflow organization of the computer room. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is one of the flow diagrams of the method for optimizing the airflow organization in a data center computer room based on an agent model provided in an embodiment of the present application;

[0034] Figure 2 This is the second flow chart of the method for optimizing the airflow organization in a data center computer room based on an agent model provided in an embodiment of the present application;

[0035] Figure 3 It is a schematic diagram of the layout of the micro-modules in the computer room provided in the embodiment of the present application;

[0036] Figure 4 It is a working condition scatter plot obtained based on the Latin hypercube sampling method provided in an embodiment of the present application;

[0037] Figure 5 It is a temperature distribution cloud diagram of a cross section of a data center room provided in an embodiment of the present application;

[0038] Figure 6 It is a velocity vector diagram of a cross section of a data center computer room provided in an embodiment of the present application;

[0039] Figure 7 It is a schematic diagram of the implementation of NSGA-II provided in the embodiment of the present application for multi-objective optimization of airflow organization;

[0040] Figure 8 It is a structural schematic diagram of an airflow organization optimization device for a data center computer room based on an agent model provided in an embodiment of the present application;

[0041] Fig. 9 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0044] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc.; multiple elements refer to two or more elements, etc.

[0045] The embodiments of the present application are described below in conjunction with the drawings in the embodiments of the present application.

[0046] Figure 1 This is one of the flow diagrams of the method for optimizing the airflow organization in a data center computer room based on an agent model provided in an embodiment of the present application, such as Figure 1 As shown, the execution subject of the method may be an electronic device, such as a server, etc. The method includes the following steps S101 to S103.

[0047] Step S101, based on each airflow organization evaluation index and the computer room airflow organization simulation data corresponding to each sample point in the design space, evaluate the computer room airflow organization simulation data, determine the airflow organization evaluation data corresponding to each sample point, and the design space is determined based on the value range corresponding to each air supply parameter.

[0048] Step S102: constructing a proxy model based on the airflow organization evaluation data corresponding to each sample point, where the proxy model is used to characterize the mapping relationship between the air supply parameters and the airflow organization evaluation index.

[0049] Step S103, based on various airflow organization evaluation indicators and the proxy model, the target air supply parameters are determined through multi-objective optimization.

[0050] The above air supply parameters may be air supply temperature, air supply speed or air supply angle.

[0051] The above-mentioned airflow organization evaluation index can be a heating index SHI, a heat mixing index HMI or an average temperature difference index β between the rack inlet and exhaust air. The heating index SHI is used to characterize the degree of mixing of the supply air with the hot air before entering the cold channel. The heat mixing index HMI is used to characterize the worst airflow mixing condition of the rack in the cabinet. The average temperature difference index β between the rack inlet and exhaust air is used to evaluate the possibility of the existence of local hot spots.

[0052] It is understandable that by constructing a proxy model (a simplified mathematical model) to approximate the real physical process, the rapid solution of airflow organization and related evaluation indicators under different operating parameters can be achieved (compared to simulating the airflow organization of the computer room through CFD simulation tools), which can effectively reduce the workload of numerical simulation and reduce the computing resources required in the optimization process. Based on various airflow organization evaluation indicators and proxy models, multi-objective optimization can be performed, which can simultaneously consider multiple design variables (air supply parameters as design variables) and target variables (airflow organization evaluation indicators as target variables) of the computer room airflow organization, avoid the single goal of the computer room airflow organization optimization, and easily find the optimal airflow organization distribution state. Therefore, by combining the technical means of constructing a proxy model with the technical means of multi-objective optimization, it is possible to achieve multiple goals of efficiently optimizing the airflow organization of the computer room.

[0053] In a possible implementation, the above-mentioned target air supply parameters are determined through multi-objective optimization based on various airflow organization evaluation indicators and proxy models, including: constructing an objective function based on various airflow organization evaluation indicators and proxy models, with the air supply parameters serving as decision variables (or design variables) of the objective function, and the airflow organization evaluation indicators serving as optimization targets (or target variables) of the objective function; based on the objective function, multi-objective optimization of the airflow organization of the computer room is performed through the second-generation non-dominated sorting genetic algorithm to determine the target air supply parameters.

[0054] In a possible implementation, before the above evaluation of the computer room airflow organization simulation data, it also includes: for each sample point, simulating the computer room airflow organization through a CFD simulation tool to obtain the computer room airflow organization simulation data corresponding to each sample point.

[0055] Optionally, for each sample point, the airflow organization of the computer room is simulated by a CFD simulation tool to obtain the simulation data of the airflow organization of the computer room corresponding to each sample point, including performing the following operations for each sample point: obtaining a computer room simulation model, the computer room simulation model is used to characterize the computer room cabinets and the ventilation method of the computer room; based on the sample points, setting the operating parameters of the computer room simulation model; based on the computer room simulation model after setting the operating parameters, simulating fluid movement by a CFD simulation tool to obtain the simulation data of the airflow organization of the computer room corresponding to the sample point.

[0056] Exemplarily, the above-mentioned acquisition of the computer room simulation model specifically includes: constructing multiple micromodules based on the cabinet simulation model and the cabinet layout method, one micromodule includes multiple cabinets, and the cabinet layout method is used to represent the positional relationship between different cabinets in the same micromodule (for example, the cabinet layout method can be to place the cabinets in two columns); constructing a computer room simulation model based on multiple micromodules and the computer room ventilation method, and the computer room ventilation method is used to characterize the distribution of cold channels and hot channels in the computer room.

[0057] The cabinet simulation model is constructed through the following steps: the upper surface, lower surface and two side surfaces of the cabinet are configured as insulating surfaces; the front and rear door panels of the cabinet are simulated by vents, and the opening rate is set; the slots in the cabinet where no servers are placed are closed with baffles; for the slots in the cabinet where servers are placed, the servers are replaced by non-physical space heat sources (blocks) with given heating power.

[0058] In order to obtain the above sample points, before simulating the airflow organization of the computer room by using the CFD simulation tool, it also includes: sampling in the design space by using the Latin Hypercube Sampling (LHS) method to obtain multiple sample points.

[0059] Figure 2This is the second flow chart of the method for optimizing the airflow organization of a data center computer room based on the proxy model provided in the embodiment of the present application, such as Figure 2 As shown, the method includes the following steps S201 to S205.

[0060] Step S201, constructing a numerical simulation model of a data center computer room.

[0061] Collect the structural dimensions and operating data of the target computer room, including the floor plan, elevation, air supply form, sensor layout, and product manuals of cabinets, precision air conditioners, and lighting equipment. Based on this, use 3D visualization modeling software and CFD simulation software to complete the construction of a numerical simulation model of the computer room.

[0062] Step S202, using Latin hypercube sampling method to obtain operating condition sample points.

[0063] According to the actual computer room environment and airflow organization requirements, determine the reasonable range of operating parameters such as air supply temperature, air supply speed and air supply angle, and select a suitable sampling method within the parameter range to generate a sample space.

[0064] Step S203, changing the boundary conditions of the numerical simulation model of the computer room according to the sampling results, controlling other parameters to remain unchanged, and performing (high-precision) simulation of the airflow organization of the computer room under different working conditions.

[0065] Step S204, selecting a suitable airflow organization evaluation index, training the sample data set obtained by numerical simulation based on the Kriging proxy model, and fitting the functional relationship between the air supply parameters and the airflow organization evaluation index.

[0066] The predictive ability of the Kriging surrogate model can be evaluated by cross-validation method.

[0067] Step S205 , constructing an objective function, and performing multi-objective optimization on the airflow organization of the computer room based on the second generation non-dominated sorting genetic algorithm.

[0068] Based on the second-generation non-dominated sorting genetic algorithm, multi-objective optimization of the airflow organization in the computer room is carried out, the objective function is constructed, and the airflow organization optimization plan is finally formulated through steps such as initializing the population, designing the fitness function, selecting the crossover mutation evolutionary population, and analyzing the Pareto optimal solution set.

[0069] The following is a detailed description of steps S201 to S205.

[0070] In step S201, taking a certain data center as an example, a numerical simulation model of a computer room (ie, the above-mentioned computer room simulation model) is constructed.

[0071] The cabinet damping model (i.e. the cabinet simulation model mentioned above) is constructed based on the Airpak simulation software. Specifically, the top, bottom and sides of the cabinet are regarded as adiabatic surfaces and the thickness is ignored. The slots (U positions) without servers in the cabinet are closed with baffles and partition simulation is adopted. For the slots with servers in the cabinet, the servers are replaced by non-solid space heat sources (blocks) with given heating power. The front and rear door panels of the cabinet are simulated by vents, and the opening rate is set to balance the internal pressure drop of the cabinet that is reduced due to simplifying the server as a non-solid space heat source.

[0072] The partition simulation is a module of the Airpak simulation software. Airpak software uses "object" as the modeling basis, and the user interface covers a variety of geometric modules, such as wall, room, vent, opening, partition, etc.

[0073] Based on the cabinet damping model, the numerical simulation model of the computer room is established. Specifically, the initial air supply mode of the IT computer room is horizontal diffuse air supply with closed heat channel. Figure 3 is a schematic diagram of the layout of the micro-modules in the computer room provided in the embodiment of the present application, such as Figure 3 As shown, the computer room has 10 groups of micro-modules, each of which contains 27 to 28 cabinets with a power of 6KW, which are placed in two rows. A hot channel is set between the two rows. The side of the cabinet that contacts the hot channel is the side where the air outlet is located, and the side where the air inlet and the side where the air outlet are located in the cabinet are opposite to each other. The ventilation method is horizontal diffuse air supply by the wind wall data processing unit (Air Handling Unit, AHU), which is delivered on both sides to diffuse the cold air to the cold channel (room cold pool). There is no raised floor. The cold channel is connected to the air inlet of the cabinet. After the cold air enters the cabinet and takes away the heat in the server cabinet, it is discharged from the cabinet outlet to the closed hot channel, and then returns to the return air outlet at the rear of the air conditioning unit through the hot ceiling.

[0074] In step S202, air supply temperature, air supply speed and air supply angle are selected as design variables, and their value ranges are set according to actual conditions. Before optimization, the air supply temperature, air supply speed and air supply angle of the data center are 22°C, 2.26m / s and 0°, respectively. Optimization is performed near this operating point, and the value ranges of the three variables are 18-25°C, 1.80-2.50m / s and -45-45°, respectively. The design space composed of design variables can be expressed as x=[T Vθ] T , select the Latin hypercube sampling method to sample in the design space, take 50 sample points for each variable, and form a 50×3 dimensional space, as shown in the following formula:

[0075]

[0076] Figure 4 is a working condition scatter diagram obtained based on the Latin hypercube sampling method provided in the embodiment of the present application, such as Figure 4 As shown, there are 150 working conditions in total, which need to be numerically simulated using the CFD model of the data center computer room established in step S201. Figure 4 The unit of air supply velocity is m / s, and the unit of air supply temperature is degree Celsius.

[0077] In step S203, the operating parameters of the numerical simulation model of the computer room are set according to the sample points obtained by sampling, including the air supply temperature, air supply speed and air supply angle. At the same time, the boundary condition type that conforms to the actual situation is set. The walls and ceiling around the computer room adopt wall insulation boundary conditions, the air conditioning supply outlet is the velocity inlet boundary condition, and the return air outlet is the pressure outlet boundary condition.

[0078] First, we assume that the airflow in the computer room is viscous and incompressible. Based on this, we establish the three major control equations of fluid motion. The results of the continuity equation, momentum equation, and energy equation are as follows:

[0079] (1) Continuity equation:

[0080]

[0081] Where ρ is the air density; t is the time variable; u, v and w are the components of the fluid velocity vector in the three directions of the coordinate system (x, y, and z axes).

[0082] (2) Momentum equation:

[0083]

[0084] In the formula, is the velocity vector; τ ij It represents the component of the viscous stress generated on the surface of the micelle due to the molecular viscosity, where i and j represent different coordinate directions, i = j is the normal stress, and i ≠ j is the shear stress; F x 、F y 、F z are the body force components in the x, y, and z directions respectively.

[0085] (3) Energy equation:

[0086]

[0087] Where, T is the thermodynamic temperature; λ is the air heat transfer coefficient; c p is the specific heat capacity; S T is the viscous dissipation term.

[0088] Then, a suitable turbulence model is selected. Since the standard k-ε model has good robustness and economy, this model is used to solve the core turbulent region. The turbulent kinetic energy k equation and the turbulent dissipation rate ε equation are as follows:

[0089]

[0090] In the formula, k represents turbulent kinetic energy, ε represents turbulent dissipation rate, ρ represents density, and u j is the component of the velocity vector in the j direction, x j is the component of the displacement vector in the j direction, μ is the air dynamic viscosity, and μ t is the turbulent viscosity (the turbulent viscosity is a function of the spatial coordinates and depends on the flow state rather than the physical parameters. The subscript t indicates turbulent flow);

[0091] G k is the term for turbulent kinetic energy caused by velocity gradient; G b is the term for turbulent kinetic energy caused by buoyancy; C1, C2, C3 are empirical constants; σ k , σ e is the turbulent Planck constant for the k equation and the ε equation.

[0092] It is understandable that the heat transfer flow process in the data center is affected and dominated by the three major conservation laws, including conservation of mass, conservation of momentum and conservation of energy. In addition to satisfying the three basic laws of flow, the flow is in a turbulent state, and the turbulent transport equations should be added, namely the turbulent kinetic energy k equation and the turbulent dissipation rate ε equation. Therefore, the basic control equations for indoor air flow and heat exchange in the data center are composed of the mass conservation equation, momentum conservation equation, energy conservation equation, k equation and ε equation.

[0093] Among the three control equations of the fluid, the convection term of the momentum equation contains nonlinear terms, the velocity component is complexly coupled in the control equation, and there is no directly solvable equation for the pressure term. Conventional solutions cannot solve the control equations. The finite volume method can be used to discretize the equations, and the interpolation method can be used to construct discrete equations. After establishing discrete equations for each variable in the flow field and determining the boundary conditions, the SIMPLE algorithm is used to solve the algebraic equations.

[0094] The SIMPLE algorithm is used to sequentially solve the discrete equations of each variable (variables in the above control equations). Except for the convection term of the momentum equation, which is discretized using the "body force weighted" method, the remaining control equations use the second-order discrete format. The relaxation factor setting remains the default, the number of iterations is set to 1000, and the convergence residual value of the momentum equation is set to 1×10 -3 , the convergence residual value of the energy equation is 1×10 -6 . Figure 5This is a temperature distribution cloud diagram of a data center room cross section provided in an embodiment of the present application. Figure 6 It is a velocity vector diagram of the cross section of the data center room provided in the embodiment of the present application, and a high-precision simulation of the airflow organization under different working conditions is performed. The (horizontal) cross section to be displayed is selected and the temperature field and velocity field distribution of the cross section are recorded, such as Figure 5 and Figure 6 As shown, it provides a basis for the calculation of airflow organization evaluation indicators. Figure 5 and Figure 6 The y value represents the distance from the horizontal plane to the floor in the vertical direction (i.e., the y-axis direction). Figure 6 The x value represents the distance from the origin of the computer room coordinate system to the vertical plane in the x-axis direction.

[0095] In step S204, a suitable airflow organization evaluation index is selected as the optimization target variable. Here, the heating index SHI, the heat mixing index HMI, and the average temperature difference index β of the rack inlet and exhaust air are taken as the optimization target variables. Among them, SHI can reflect the degree of mixing of the supply air with the hot air before entering the cold channel. The closer the index is to 0, the less mixing of cold and hot air flows, and the higher the air supply efficiency of the air conditioner.

[0096]

[0097] Where δQ is the heat absorbed by the cold air flow from the floor air outlet to the air inlet of the server cabinet; Q is the cooling loss of the cold air flow in the cabinet for cooling the server.

[0098] HMI can reflect the worst airflow mixing situation of the racks in the cabinet, and can also make a certain assessment of the overall temperature environment of the racks in the cabinet. The smaller the index is, the less the degree of mixing of cold and hot airflows is, and the more stable the thermal environment of the data center is.

[0099]

[0100] Where, T in 、T out is the average temperature of the rack’s inlet and outlet air; T in-max 、T in-min is the highest and lowest temperatures of the rack air inlet section; T ref is the reference temperature (air conditioning supply air temperature).

[0101] β can evaluate the possibility of the existence of local hot spots. The smaller the index is, the higher the airflow organization evaluation is. If it is greater than 1, it indicates that exhaust self-circulation has occurred in the rack. The target value is 0.

[0102]

[0103] Then, the sample data set obtained by numerical simulation was divided, and stratified sampling method was used to select 70% training set and 30% test set. The data set was trained based on Kriging proxy model, and the functional relationship between air supply parameters and airflow organization evaluation index was fitted. Finally, the prediction ability of the model was evaluated by cross-validation method, where the Kirging model can be expressed as:

[0104] y(x)=f T (x)β+z(x);

[0105] In the formula, f T (x) represents a low-precision analysis model, which can be a low-precision simulation model or a global approximation model constructed using a large number of low-precision sample points; the proportional function β represents the global trend function of the position; Z(x) is a stationary random process with zero mean; y(x) is the response result of the Kirging model.

[0106] Figure 7 Schematic diagram of the implementation of NSGA-II for multi-objective optimization of airflow organization provided in the embodiment of the present application, such as Figure 7 As shown, in step S205, multi-objective optimization is performed on the airflow organization of the computer room based on the second generation non-dominated sorting genetic algorithm (NSGA-II). First, the objective function is constructed according to the functional relationship between the air supply parameters and the airflow organization evaluation index obtained by fitting. Then, chromosome encoding is performed to encode the operating parameters of the computer room into chromosomes. Each gene represents a parameter, and the overall structure of the chromosome represents a possible solution. Then, the population is initialized and a set of initial solutions, namely the initial population, is randomly generated. These solutions will serve as the starting point of the optimization process. Fitness evaluation uses the objective function to evaluate the fitness of each individual. Fitness is the basis of natural selection in genetic algorithms and determines the probability that an individual is selected to participate in reproduction. According to the fitness, the better individuals are selected from the current population. Roulette selection, tournament selection and other methods can be used. The genes of the selected parent individuals are combined through crossover (or hybridization) operations to generate new offspring. Mutation operation randomly changes certain genes of certain individuals with a certain probability to increase the diversity of the population and prevent premature convergence. The fitness of the newly generated offspring is evaluated and integrated into the population to keep the population size unchanged. If the termination condition is met, such as reaching the preset number of iterations or the quality of the solution has met the requirements, the algorithm is stopped. Otherwise, the fitness evaluation step is returned to continue iteration. Finally, the optimized airflow organization scheme and its performance indicators are output.

[0107] The following is a description of the data center computer room airflow organization optimization device based on the proxy model provided in the present application. The data center computer room airflow organization optimization device based on the proxy model described below and the data center computer room airflow organization optimization method based on the proxy model described above can be referenced to each other.

[0108] Figure 8 is a structural schematic diagram of a data center computer room airflow organization optimization device based on an agent model provided in an embodiment of the present application, such as Figure 8 As shown, the device includes: a simulation data evaluation module 10, an agent model construction module 20 and a multi-objective optimization module 30.

[0109] A simulation data evaluation module 10 is used to evaluate the airflow organization simulation data of the computer room based on each airflow organization evaluation index and the airflow organization simulation data of the computer room corresponding to each sample point in the design space, and determine the airflow organization evaluation data corresponding to each sample point, wherein the design space is determined based on the value range corresponding to each air supply parameter;

[0110] An agent model building module 20 is used to build an agent model based on the airflow organization evaluation data corresponding to each sample point, and the agent model is used to characterize the mapping relationship between the air supply parameters and the airflow organization evaluation index;

[0111] The multi-objective optimization module 30 is used to determine the target air supply parameters through multi-objective optimization based on various airflow organization evaluation indicators and proxy models.

[0112] It can be understood that the detailed functional implementation of each of the above-mentioned units / modules can be found in the introduction of the aforementioned method embodiment, and will not be repeated here.

[0113] It should be understood that the above-mentioned device is used to execute the method in the above-mentioned embodiment. The implementation principle and technical effect of the corresponding program module in the device are similar to those described in the above-mentioned method. The working process of the device can refer to the corresponding process in the above-mentioned method, which will not be repeated here.

[0114] Based on the method in the above embodiment, an embodiment of the present application provides an electronic device, Fig. 9 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Fig. 9 As shown, the electronic device may include: a processor (Processor) 810, a communication interface (Communications Interface) 820, a memory (Memory) 830 and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call the logic instructions in the memory 830 to execute the method in the above embodiment.

[0115] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application.

[0116] Based on the method in the above embodiment, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method in the above embodiment.

[0117] Based on the method in the above embodiment, an embodiment of the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method in the above embodiment.

[0118] It is understandable that the processor in the embodiment of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0119] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.

[0120] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions may be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

[0121] It should be understood that the various numerical numbers involved in the embodiments of the present application are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application.

[0122] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for optimizing airflow organization in a data center computer room based on an agent model, characterized in that: include: Based on each airflow organization evaluation index and the computer room airflow organization simulation data corresponding to each sample point in the design space, the computer room airflow organization simulation data is evaluated to determine the airflow organization evaluation data corresponding to each sample point, wherein the design space is determined based on the value range corresponding to each air supply parameter; Based on the airflow organization evaluation data corresponding to each sample point, a proxy model is constructed, wherein the proxy model is used to characterize the mapping relationship between the air supply parameters and the airflow organization evaluation index; Based on various airflow organization evaluation indicators and the proxy model, target air supply parameters are determined through multi-objective optimization.

2. The method for optimizing airflow organization in a data center computer room based on an agent model according to claim 1, characterized in that: The target air supply parameters are determined through multi-objective optimization based on various airflow organization evaluation indicators and the proxy model, including: Based on various airflow organization evaluation indicators and the proxy model, an objective function is constructed, the air supply parameter is used as a decision variable of the objective function, and the airflow organization evaluation indicator is used as an optimization target of the objective function; Based on the objective function, the multi-objective optimization of the airflow organization in the computer room is performed through the second generation non-dominated sorting genetic algorithm to determine the target air supply parameters.

3. The method for optimizing airflow organization in a data center computer room based on an agent model according to claim 1, characterized in that: Before evaluating the airflow simulation data of the computer room, it also includes: For each sample point, the airflow organization of the computer room is simulated by using the computational fluid dynamics (CFD) simulation tool to obtain the simulation data of the airflow organization of the computer room corresponding to each sample point.

4. The method for optimizing airflow organization in a data center computer room based on an agent model according to claim 3 is characterized in that: For each sample point, the airflow organization of the computer room is simulated by a computational fluid dynamics (CFD) simulation tool to obtain the simulation data of the airflow organization of the computer room corresponding to each sample point, including performing the following operations for each sample point: Obtain a computer room simulation model, which is used to characterize computer room cabinets and ventilation methods; Based on the sample points, setting the operating parameters of the computer room simulation model; Based on the computer room simulation model after setting the working condition parameters, the fluid movement is simulated by a computational fluid dynamics (CFD) simulation tool to obtain the computer room airflow organization simulation data corresponding to the sample point.

5. The method for optimizing airflow organization in a data center computer room based on an agent model according to claim 3 is characterized in that: Before simulating the airflow organization of the computer room through computational fluid dynamics (CFD) simulation tools, it also includes: Sampling is performed in the design space using the Latin hypercube sampling method to obtain multiple sample points.

6. The method for optimizing airflow organization in a data center computer room based on an agent model according to any one of claims 1 to 5, characterized in that: The air supply parameter is air supply temperature, air supply speed or air supply angle.

7. The method for optimizing airflow organization in a data center computer room based on an agent model according to any one of claims 1 to 5, characterized in that: The airflow organization evaluation index is a heating index SHI, a heat mixing index HMI or a rack inlet and exhaust air average temperature difference index β; The heating index SHI is used to characterize the degree of mixing of the supply air with the hot air before entering the cold aisle; The thermal mixing index HMI is used to characterize the worst airflow mixing condition of the rack in the cabinet; The rack inlet and outlet air average temperature difference index β is used to evaluate the possibility of the existence of local hot spots.

8. A data center computer room airflow organization optimization device based on an agent model, characterized in that: include: A simulation data evaluation module is used to evaluate the airflow organization simulation data of the computer room based on each airflow organization evaluation index and the airflow organization simulation data of the computer room corresponding to each sample point in the design space, and determine the airflow organization evaluation data corresponding to each sample point, wherein the design space is determined based on the value range corresponding to each air supply parameter; An agent model building module is used to build an agent model based on the airflow organization evaluation data corresponding to each sample point, and the agent model is used to characterize the mapping relationship between the air supply parameters and the airflow organization evaluation index; The multi-objective optimization module is used to determine the target air supply parameters through multi-objective optimization based on various airflow organization evaluation indicators and the proxy model.

9. An electronic device, characterized in that: include: at least one memory for storing a computer program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program runs on a processor, the processor is caused to execute the method according to any one of claims 1 to 7.

Citation Information

Cited By

  • Indoor ventilation airflow intelligent design method oriented to multiple requirements

    CN120633012A

  • Dynamic control method for wind-liquid collaborative cooling tail end of data center under load fluctuation

    CN121920243A

  • Temperature and airflow prediction system and method thereof

    TWI909918B