Self-adaptive calibration method and device for intelligent data center
By using pre-trained neural network proxy model in the data center to calibrate the server air flow rate, combined with the CFD model, the problems of insufficient accuracy and high computing cost in data center simulation are solved, and more efficient temperature and air velocity distribution map prediction is achieved. .
Patent Information
- Application Number
- CN202411229801.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-11-22
AI Technical Summary
In the prior art, when data centers use digital twins for simulation and optimization, due to parameter uncertainty and complexity of simulation calculation, insufficient accuracy and high calculation costs.
Adaptive calibration method for intelligent data centers is adopted to calibrate the server air flow rate through a pre-trained neural network proxy model. Combined with softmax activation function and CFD model, the solution frequency of the CFD model is reduced, the calculation cost is reduced, and the model accuracy is improved.
It significantly improves the accuracy of the data center temperature and air velocity distribution map, reduces the calculation cost, reduces the solution frequency of the CFD model, and improves the accuracy of the model.
Smart Images

Figure CN119012646A_ABST
Abstract
Description
Technical Field
[0001] The present specification relates to the fields of computer technology and cooling control technology, and in particular to an adaptive calibration method and device for an intelligent data center. Background Art
[0002] As an important part of modern information technology infrastructure, data centers are increasing in size and complexity with the rapid development of new technologies such as cloud computing, big data, and big models. When data center servers and equipment are running, they will generate a lot of heat. If it cannot be dissipated in time, local hot spots will be generated, resulting in server performance degradation and shortened equipment life, which will in turn affect the stability and security of the data center. Therefore, the simulation and optimization of data centers are particularly important.
[0003] In the existing technology, data centers use digital twins to simulate and optimize their operations, which usually involves using physical modeling tools to create detailed simulations of data center systems. However, these models may suffer from insufficient accuracy and high computational costs due to parameter uncertainty and the complexity of simulation calculations.
[0004] Based on this, an embodiment of this specification provides an adaptive calibration method for an intelligent data center. Summary of the invention
[0005] The embodiments of this specification provide an adaptive calibration method and device for an intelligent data center, which is used to solve the following technical problems: In the prior art, data centers use digital twins to simulate and optimize their operations, which usually involves using physical modeling tools to create detailed simulations of data center systems. However, these models may have insufficient accuracy and high computational costs due to parameter uncertainty and complexity of simulation calculations.
[0006] To solve the above technical problems, the embodiments of this specification are implemented as follows:
[0007] The embodiment of this specification provides an adaptive calibration method for an intelligent data center, including:
[0008] Get the current status of the smart data center to be processed;
[0009] Based on the pre-trained neural network agent model, the server air flow rate of the intelligent data center to be processed is calibrated to obtain a calibrated server air flow rate, wherein the preset neural agent model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of the CRAC set point and the fan speed on the temperature of all sensor positions, and the cooling block uses a softmax activation function to calculate a cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and the air flow rate;
[0010] Based on the calibrated server air flow rate, calibrating a preset CFD model to obtain a calibrated CFD model;
[0011] The current state of the intelligent data center to be processed is used as an input of the calibrated CFD model to obtain a temperature and air velocity distribution diagram of the intelligent data center to be processed.
[0012] The embodiment of this specification also provides an adaptive calibration device for an intelligent data center, including:
[0013] An acquisition module obtains the current status of the intelligent data center to be processed;
[0014] An air flow rate calibration module, based on a pre-trained neural network proxy model, calibrates the server air flow rate of the intelligent data center to be processed to obtain a calibrated server air flow rate, wherein the preset neural proxy model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of CRAC set point and fan speed on the temperature of all sensor positions, the cooling block uses a softmax activation function to calculate a cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and air flow rate;
[0015] A model calibration module, based on the calibrated server air flow rate, calibrates a preset CFD model to obtain a calibrated CFD model;
[0016] The processing module uses the current state of the intelligent data center to be processed as an input of the calibrated CFD model to obtain a temperature and air velocity distribution diagram of the intelligent data center to be processed.
[0017] An adaptive calibration method for an intelligent data center provided in an embodiment of the present specification obtains the current state of the intelligent data center to be processed; based on a pre-trained neural network agent model, calibrates the server air flow rate of the intelligent data center to be processed to obtain a calibrated server air flow rate, wherein the preset neural agent model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of CRAC set point and fan speed on the temperature of all sensor positions, and the cooling block uses a softmax activation function to calculate a cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and air flow rate; based on the calibrated server air flow rate, calibrates the preset CFD model to obtain a calibrated CFD model; uses the current state of the intelligent data center to be processed as the input of the calibrated CFD model to obtain the temperature and air velocity distribution diagram of the intelligent data center to be processed, which can reduce the solution frequency of the CFD model, reduce the calculation cost, and significantly improve the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0019] Figure 1 A schematic diagram of the system architecture of an adaptive calibration method for an intelligent data center provided in an embodiment of this specification;
[0020] Figure 2 A flowchart of an adaptive calibration method for an intelligent data center provided in an embodiment of this specification;
[0021] Figure 3 A schematic diagram of an adaptive calibration device for an intelligent data center provided in an embodiment of this specification. DETAILED DESCRIPTION
[0022] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.
[0023] Figure 1 The following is a schematic diagram of the system architecture of an adaptive calibration method for an intelligent data center provided in an embodiment of this specification. Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0024] The terminal devices 101, 102, 103 interact with the server 105 via the network 104 to receive or send messages, etc. Various client applications may be installed on the terminal devices 101, 102, 103, such as dedicated programs such as adaptive calibration methods for intelligent data centers.
[0025] The terminal devices 101, 102, 103 may be hardware or software. When the terminal devices 101, 102, 103 are hardware, they may be various dedicated or general electronic devices, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, etc. When the terminal devices 101, 102, 103 are software, they may be installed in the electronic devices listed above. They may be implemented as multiple software or software modules (e.g., multiple software or software modules for providing distributed services), or they may be implemented as a single software or software module.
[0026] The server 105 may be a server that provides various services, such as a backend server that provides services for client applications installed on the terminal devices 101, 102, and 103. For example, the server may perform adaptive calibration for the intelligent data center so as to display the adaptive calibration result for the intelligent data center on the terminal device servers 101, 102, and 103, and the server may also perform adaptive calibration based on the intelligent data center so as to display the adaptive calibration result based on the intelligent data center on the terminal devices 101, 102, and 103.
[0027] The server 105 may be hardware or software. When the server 105 is hardware, it may be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server 105 is software, it may be implemented as multiple software or software modules (e.g., multiple software or software modules for providing distributed services), or as a single software or software module.
[0028] Figure 2The following is a flowchart of an adaptive calibration method for an intelligent data center provided in an embodiment of the present specification. From a program perspective, the execution subject of the process can be a program installed on an application server or an application terminal. It can be understood that the method can be executed by any device, equipment, platform, or device cluster with computing and processing capabilities. Figure 2 As shown, the adaptive calibration method includes:
[0029] Step S201: Acquire the current status of the intelligent data center to be processed.
[0030] In the embodiment of the present specification, the intelligent data center to be processed includes 1 CRAC unit, m servers and n temperature sensors deployed in the cold and hot channels.
[0031] In the embodiment of this specification, the intelligent data center to be processed is an intelligent data center that requires cooling process control.
[0032] In the embodiment of the present specification, the current state of the intelligent data center to be processed includes: the server air flow rate of the intelligent data center to be processed and the boundary conditions of the intelligent data center to be processed. Specifically, the boundary conditions of the intelligent data center to be processed include: the CRAC set point T c , CRAC fan speed V and server power P.
[0033] Step S203: Based on the pre-trained neural network agent model, the server air flow rate of the intelligent data center to be processed is calibrated to obtain a calibrated server air flow rate, wherein the preset neural agent model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of CRAC set point and fan speed on the temperature of all sensor positions, and the cooling block uses a softmax activation function to calculate the cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and air flow rate.
[0034] In the embodiments of this specification, the role of the pre-trained neural network proxy model is to avoid directly solving the optimization problem of the preset CFD model to reduce the computing cost.
[0035] In the embodiment of this specification, the training of the pre-trained neural network proxy model includes:
[0036] A training data set is generated based on a preset CFD model and an initialization parameter set, wherein the initialization parameters are the preset initial parameters of the data center, and the initialization parameters include: CRAC set point T c , fan speed V, server power P and server air flow rate α;
[0037] Based on the training data set, the weight of the neural network proxy model is adjusted by minimizing the difference between the temperature predicted by the neural network proxy model and the temperature predicted by the preset CFD model to obtain the pre-trained neural network proxy model, wherein the minimization of the difference between the temperature predicted by the neural network proxy model and the temperature predicted by the preset CFD model is specifically:
[0038]
[0039] in,
[0040] W * Represents the weight of the adjusted neural network agent model;
[0041] W represents the weight of the neural network proxy model;
[0042] x represents input data;
[0043] represents the predicted temperature of the neural network agent model;
[0044] Indicates the predicted temperature of the preset CFD model.
[0045] In the embodiment of this specification, the pre-trained neural network proxy model further includes:
[0046] A regularization term is added to penalize the overcooling or overheating conditions to adjust the server air flow rate α to obtain a calibrated server air flow rate, thereby minimizing the difference between the pre-trained neural network agent model and the actual temperature of the intelligent data center to be processed. The calibrated server air flow rate is:
[0047]
[0048] in,
[0049] α * is the calibrated server air flow rate;
[0050] represents the predicted temperature of the neural network agent model;
[0051] represents the predicted temperature of the preset CFD model;
[0052] λ represents the weight parameter of the regularization term;
[0053] k represents the index of the sensor;
[0054] h(T) means adding a penalty for temperature difference to the loss function.
[0055] λ is used to control the influence of the regularization term in the overall loss function. By adjusting the value of λ, the relationship between the model's fitting accuracy and physical constraints can be balanced.
[0056] k ranges from 1 to n, where n is the total number of sensors.
[0057] h(T) is a penalty function used to introduce physical constraints in the calibration process of the proxy model, which can ensure that the temperature gradient predicted by the model is within a reasonable empirical range.
[0058] In an embodiment of the present specification, based on a pre-trained neural network proxy model, by adjusting the server air flow rate to minimize the difference in actual temperature between the pre-trained neural network proxy model and the intelligent data center to be processed, the consistency between the predicted value of the pre-trained neural network proxy model and the actual sensor of the intelligent data center to be processed can be ensured, thereby ensuring the prediction accuracy of the pre-trained neural network proxy model.
[0059] Step S205: Based on the calibrated server air flow rate, calibrate the preset CFD model to obtain a calibrated CFD model.
[0060] In the embodiment of the present specification, the calibrating the preset CFD model based on the calibrated server air flow rate to obtain the calibrated CFD model specifically includes:
[0061] Using the calibrated server air flow rate, configuring the preset CFD model to obtain a configured CFD model;
[0062] If the prediction accuracy of the configured CFD model meets a preset target, or the generation of the configured CFD model reaches a maximum number of iterations, the calibrated CFD model is obtained.
[0063] In the embodiment of this specification, the maximum number of iterations may be 50. The preset target may be 95%.
[0064] In the embodiments of the present specification, the preset CFD model is obtained by adopting a model simplification technology based on proper orthogonal decomposition (POD). The temperature field is approximated as a linear combination of a set of spatial basis functions (i.e., POD modes) to describe the profile of the entire field. This method not only reduces the demand for computing resources, but also shortens the response time, thereby accelerating the simulation speed.
[0065] In the embodiment of this specification, obtaining the preset CFD model includes:
[0066] Based on the CFD simulation, CFD training samples are generated using the calibrated server air flow rate and other preset boundary conditions, wherein the other preset boundary conditions include: CRAC set point T c , fan speed V, server power P;
[0067] Based on the correlation matrix C determined by the CFD training sample, solving the eigenvector of the correlation matrix C to determine the POD mode;
[0068] Based on the POD model, construct a linear equation;
[0069] Solving the linear equation using the least square method to obtain the boundary coefficient corresponding to the POD mode;
[0070] Based on the POD mode and the boundary coefficient corresponding to the POD mode, temperature prediction is performed and the preset CFD model is updated.
[0071] In the embodiment of this specification, the correlation matrix C is:
[0072] C ij = <T(x i ),T(x j )>,
[0073] in,
[0074] T(x i ) indicates that the preset CFD model is i Temperature field output from sample points;
[0075] T(x j ) indicates that the preset CFD model is j Temperature field output from sample points;
[0076] <·,·> represents the inner product of two vectors;
[0077] The expression for solving the eigenvector of the correlation matrix C is:
[0078]
[0079] in,
[0080] λ j Representation and eigenvectors;
[0081] Φ j represents the relevant eigenvalue;
[0082] M represents the number of samples collected;
[0083] T(x i ) indicates that the CFD model is in xi Temperature field output from sample points;
[0084] T(x i ) T Represents the temperature field T(x i )’s transposed vector;
[0085] The linear equation is expressed as:
[0086]
[0087] in,
[0088] β i (x) represents the coefficient of the i-th POD mode;
[0089] represents the value of the i-th POD mode associated with the first server at the air outlet;
[0090] represents the value of the i-th POD mode associated with the first server at the air inlet;
[0091] It indicates that this is the value of the i-th POD mode associated with the second server at the air outlet; It indicates that this is the value of the i-th POD mode associated with the second server at the air inlet; represents the value of the i-th POD mode associated with the k-th CRAC unit at the return air; V represents the value of the i-th POD mode associated with the k-th CRAC unit at the air supply; k represents the fan speed of the kth CRAC unit;
[0092] P1 represents the power consumption of the first server;
[0093] P2 represents the power consumption of the second server;
[0094] P m represents the power consumption of the mth server;
[0095] P j represents the power consumption of the jth server;
[0096] c p represents the specific heat capacity of air;
[0097] ρ represents the density of air;
[0098] α represents the air flow rate flowing through the jth server;
[0099] The boundary coefficient corresponding to the POD mode is:
[0100]
[0101] in,
[0102] β * Represents the optimal POD coefficient;
[0103] φ i Indicates the i-th POD mode;
[0104] T(x) represents the temperature field output by the preset CFD model.
[0105] In the embodiments of this specification, β i (x) represents the contribution of this mode to the overall temperature field under given boundary conditions x; represents the value of the i-th POD mode associated with the first server at the air outlet, indicating the thermal energy state of the air after passing through the server; represents the value of the i-th POD mode associated with the first server at the air inlet, indicating the thermal energy state of the air after passing through the server; Indicates the thermal state of the air as it returns to the CRAC unit; Indicates the thermal energy state when the CRAC unit supplies air; P1 indicates the power consumed by the first server during operation, which is eventually released in the form of heat; P2 indicates the power consumed by the second server during operation, which is eventually released in the form of heat; c p It represents the energy required to raise the temperature of air by one degree per unit mass, which is an important physical quantity to describe the thermal performance of air; ρ represents the mass of air per unit volume. In energy balance calculation, density is used to determine the mass flow of air; α represents the air flow through the jth server, which has a direct impact on the heat transfer of the air inside the server.
[0106] In the embodiment of this specification, the temperature prediction is performed based on the POD mode and the boundary coefficient corresponding to the POD mode, and the preset CFD model is updated, which specifically includes:
[0107] The POD mode and the boundary coefficient corresponding to the POD mode, and the average temperature of the CFD training sample are linearly combined to reconstruct the temperature field for temperature prediction, and update the preset CFD model;
[0108] In the embodiment of this specification, the expression for temperature field reconstruction is:
[0109]
[0110] T meanrepresents the average temperature of the CFD training sample;
[0111] β i (x) represents the boundary coefficient corresponding to the i-th POD mode;
[0112] φ i Represents the i-th POD mode.
[0113] Step S207: taking the current state of the intelligent data center to be processed as the input of the calibrated CFD model to obtain the temperature and air velocity distribution diagram of the intelligent data center to be processed.
[0114] Since the calibrated CFD model can determine the server air flow rate configuration, so that the error between the model output and the actual measurement value is minimized, the current state of the intelligent data center to be processed is used as the input data x = (T c ,V,P,α), where T c is the CRAC set point, V is the CRAC fan speed, P is the server power, and α is the server air flow rate. The temperature and air velocity distribution map of the intelligent data center to be processed can be obtained.
[0115] The adaptive calibration method for intelligent data centers provided in the embodiments of the present specification obtains the current state of the intelligent data center to be processed; based on the pre-trained neural network agent model, calibrates the server air flow rate of the intelligent data center to be processed to obtain a calibrated server air flow rate, wherein the preset neural agent model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of CRAC set point and fan speed on the temperature of all sensor positions, and the cooling block uses a softmax activation function to calculate the cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and air flow rate; based on the calibrated server air flow rate, calibrates the preset CFD model to obtain a calibrated CFD model; uses the current state of the intelligent data center to be processed as the input of the calibrated CFD model to obtain the temperature and air velocity distribution diagram of the intelligent data center to be processed, which can reduce the solution frequency of the CFD model, reduce the calculation cost, and can significantly improve the accuracy of the model.
[0116] The above content describes in detail an adaptive calibration method for an intelligent data center. Correspondingly, this specification also provides an adaptive calibration device for an intelligent data center, such as Figure 3 shown. Figure 3 A schematic diagram of an adaptive calibration device for an intelligent data center provided in an embodiment of this specification, the adaptive calibration device comprising:
[0117] An acquisition module 301 acquires the current state of the intelligent data center to be processed;
[0118] The air flow rate calibration module 303 calibrates the air flow rate of the server of the intelligent data center to be processed based on the pre-trained neural network agent model to obtain the calibrated server air flow rate, wherein the preset neural agent model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of the CRAC set point and the fan speed on the temperature of all sensor positions, and the cooling block uses the softmax activation function to calculate the cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and the air flow rate;
[0119] A model calibration module 305 calibrates a preset CFD model based on the calibrated server air flow rate to obtain a calibrated CFD model;
[0120] The processing module 307 uses the current state of the intelligent data center to be processed as an input of the calibrated CFD model to obtain a temperature and air velocity distribution diagram of the intelligent data center to be processed.
[0121] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0122] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, electronic device, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.
[0123] The apparatus, electronic device, non-volatile computer storage medium and method provided in the embodiments of this specification correspond to each other, and therefore, the apparatus, electronic device and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, electronic device and non-volatile computer storage medium will not be repeated here.
[0124] In the 1990s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.
[0125] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.
[0126] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0127] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0128] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may be in the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the embodiments of this specification may be in the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.
[0129] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0130] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0131] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0132] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0133] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0134] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0135] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0136] The specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media including storage devices.
[0137] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0138] The above is only an embodiment of this specification and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims
1. An adaptive calibration method for an intelligent data center, characterized in that: The adaptive calibration method comprises: Get the current status of the smart data center to be processed; Based on the pre-trained neural network agent model, the server air flow rate of the intelligent data center to be processed is calibrated to obtain a calibrated server air flow rate, wherein the preset neural agent model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of the CRAC set point and the fan speed on the temperature of all sensor positions, and the cooling block uses a softmax activation function to calculate a cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and the air flow rate; Based on the calibrated server air flow rate, calibrating a preset CFD model to obtain a calibrated CFD model; The current state of the intelligent data center to be processed is used as an input of the calibrated CFD model to obtain a temperature and air velocity distribution diagram of the intelligent data center to be processed.
2. The adaptive calibration method according to claim 1, characterized in that: The intelligent data center to be processed includes 1 CRAC unit, m servers and n temperature sensors deployed in cold and hot channels.
3. The adaptive calibration method according to claim 1, wherein: The training of the pre-trained neural network proxy model includes: A training data set is generated based on a preset CFD model and an initialization parameter set, wherein the initialization parameters are the preset initial parameters of the data center, and the initialization parameters include: CRAC set point T c , fan speed V, server power P and server air flow rate α; Based on the training data set, the weight of the neural network proxy model is adjusted by minimizing the difference between the temperature predicted by the neural network proxy model and the temperature predicted by the preset CFD model to obtain the pre-trained neural network proxy model, wherein the minimization of the difference between the temperature predicted by the neural network proxy model and the temperature predicted by the preset CFD model is specifically: in, W * Represents the weight of the adjusted neural network agent model; W represents the weight of the neural network proxy model; x represents input data; represents the predicted temperature of the neural network agent model; Indicates the predicted temperature of the preset CFD model.
4. The adaptive calibration method according to claim 3, characterized in that: The pre-trained neural network proxy model further includes: A regularization term is added to penalize the overcooling or overheating conditions to adjust the server air flow rate α to obtain a calibrated server air flow rate, thereby minimizing the difference between the pre-trained neural network agent model and the actual temperature of the intelligent data center to be processed. The calibrated server air flow rate is: in, α * is the calibrated server air flow rate; represents the predicted temperature of the neural network agent model; represents the predicted temperature of the preset CFD model; λ represents the weight parameter of the regularization term; k represents the index of the sensor; h(T) means adding a penalty for temperature difference to the loss function.
5. The adaptive calibration method according to claim 1, wherein: The step of calibrating a preset CFD model based on the calibrated server air flow rate to obtain a calibrated CFD model specifically includes: Using the calibrated server air flow rate, configuring the preset CFD model to obtain a configured CFD model; If the prediction accuracy of the configured CFD model meets a preset target, or the generation of the configured CFD model reaches a maximum number of iterations, the calibrated CFD model is obtained.
6. The adaptive calibration method according to claim 5, characterized in that: The acquisition of the preset CFD model includes: Based on the CFD simulation, CFD training samples are generated using the calibrated server air flow rate and other preset boundary conditions, wherein the other preset boundary conditions include: CRAC set point T c , fan speed V, server power P; Based on the correlation matrix C determined by the CFD training sample, solving the eigenvector of the correlation matrix C to determine the POD mode; Based on the POD model, construct a linear equation; Solving the linear equation using the least square method to obtain the boundary coefficient corresponding to the POD mode; Based on the POD mode and the boundary coefficient corresponding to the POD mode, temperature prediction is performed and the preset CFD model is updated.
7. The adaptive calibration method according to claim 6, characterized in that: The correlation matrix C is: C ij =<T(x i ),T(x j )>, in, T(x i ) indicates that the preset CFD model is i Temperature field output from sample points; T(x j ) indicates that the preset CFD model is j Temperature field output from sample points; <·,·> represents the inner product of two vectors; The expression for solving the eigenvector of the correlation matrix C is: in, λ j Representation and eigenvectors; Φ j represents the relevant eigenvalue; M represents the number of samples collected; T(x i ) indicates that the CFD model is in x i Temperature field output from sample points; T(x i ) T Represents the temperature field T(x i )’s transposed vector; The linear equation is expressed as: in, β i (x) represents the coefficient of the i-th POD mode; represents the value of the i-th POD mode associated with the first server at the air outlet; represents the value of the i-th POD mode associated with the first server at the air inlet; It indicates that this is the value of the i-th POD mode associated with the second server at the air outlet; It indicates that this is the value of the i-th POD mode associated with the second server at the air inlet; represents the value of the i-th POD mode associated with the k-th CRAC unit at the return air; V represents the value of the i-th POD mode associated with the k-th CRAC unit at the air supply; k represents the fan speed of the kth CRAC unit; P1 represents the power consumption of the first server; P2 represents the power consumption of the second server; P m represents the power consumption of the mth server; P j represents the power consumption of the jth server; c p represents the specific heat capacity of air; ρ represents the density of air; α represents the air flow rate flowing through the jth server; The boundary coefficient corresponding to the POD mode is: in, β * Represents the optimal POD coefficient; φ i represents the i-th POD mode; T(x) represents the temperature field output by the preset CFD model.
8. The adaptive calibration method according to claim 6, wherein: The step of performing temperature prediction based on the POD mode and the boundary coefficient corresponding to the POD mode and updating the preset CFD model specifically includes: The POD mode and the boundary coefficient corresponding to the POD mode, and the average temperature of the CFD training sample are linearly combined to reconstruct the temperature field for temperature prediction, and update the preset CFD model.
9. The adaptive calibration method according to claim 8, characterized in that: The expression for temperature field reconstruction is: T mean represents the average temperature of the CFD training sample; β i (x) represents the boundary coefficient corresponding to the i-th POD mode; φ i Represents the i-th POD mode.
10. An adaptive calibration device for an intelligent data center, characterized in that: The adaptive calibration device comprises: An acquisition module obtains the current status of the intelligent data center to be processed; An air flow rate calibration module, based on a pre-trained neural network proxy model, calibrates the server air flow rate of the intelligent data center to be processed to obtain a calibrated server air flow rate, wherein the preset neural proxy model includes: a cooling block and a heating block, the cooling block is used to simulate the influence of CRAC set point and fan speed on the temperature of all sensor positions, the cooling block uses a softmax activation function to calculate a cooling coefficient matrix; the heating block is used to simulate the influence of the server on the temperature of the hot channel sensor position, and the heating block predicts the temperature increase caused by the server based on the server power and air flow rate; A model calibration module, based on the calibrated server air flow rate, calibrates a preset CFD model to obtain a calibrated CFD model; The processing module uses the current state of the intelligent data center to be processed as an input of the calibrated CFD model to obtain a temperature and air velocity distribution diagram of the intelligent data center to be processed.
