Refrigerating system energy consumption optimization method and system for computing center and medium

Abstract

The invention discloses a refrigeration system energy consumption optimization method and system for a computing center and a medium. The refrigeration system energy consumption optimization method for the computing center comprises the steps that a regulation environment needing to be started by a refrigeration system and regulation factors of the regulation environment are configured; measuring the thermal state of the host system of the computing center and the environmental information of the regulation and control environment; and a pre-trained refrigeration control decision model is adopted to decide a control action on a regulation factor of the refrigeration system based on the heat state of the host system and the environment information of the regulation environment. The invention aims to carry out self-adaptive control and adjustment on the refrigerating system according to the physical thermal state of the host system in the large-scale computing center, realize real-time, refined and intelligent regulation and control of the energy consumption of the refrigerating system, and effectively meet the requirements of energy conservation and emission reduction.

Description

Technical Field

[0001] The present invention belongs to the technical field of refrigeration energy consumption optimization, and particularly relates to a method, a system and a medium for optimizing the energy consumption of a refrigeration system for a computing center. Background Art

[0002] With the rapid development of the digitalization process of the whole society, the demand for computing power in all walks of life is very urgent. As the main carrier of computing power, large-scale computing centers have become important infrastructure for the digital economy. The construction of large-scale computing centers such as data centers and supercomputing centers has entered a new stage of high-speed development, further accelerating the industrialization and large-scale application of digital technologies.

[0003] While the host system of a large-scale computing center provides powerful computing power, it also continuously generates a large amount of heat. In order to enable each computing power component to work within a safe temperature range and exert its best performance, the refrigeration system needs to provide sufficient cooling capacity. Currently, large-scale computing centers often adopt a relatively crude refrigeration control method, resulting in significant waste of refrigeration energy consumption. Relevant research shows that the refrigeration system has become the subsystem with the highest energy consumption ratio except for the host system in large-scale computing centers. From the perspective of green energy conservation, it is urgent to optimize the energy consumption of the refrigeration system in large-scale computing centers.

[0004] Currently, there have been a lot of studies on the energy consumption optimization of refrigeration systems, mainly focusing on the energy consumption modeling analysis of refrigeration systems. Based on the quantitative analysis results of the correlation between system energy consumption and the key parameters of each device, suggestions are put forward for optimizing the operating parameters of the refrigeration system, thereby reducing the energy consumption of the refrigeration system. However, the overall structure of the refrigeration system in a large-scale computing center is complex, with a wide variety of functional devices, large quantity scales, and characteristics such as multi-level and mutual constraints, resulting in complex non-linear relationships between the operating parameters of the refrigeration system and the cooling capacity and refrigeration energy consumption. Moreover, the operating conditions of refrigeration equipment may change over time, and it is difficult for existing technologies to formulate clear control strategies according to actual refrigeration requirements. Summary of the Invention

[0005] The technical problem to be solved by the present invention: Aiming at the above problems of the prior art, the present invention provides a method, a system and a medium for optimizing the energy consumption of a refrigeration system for a computing center. The present invention aims to adaptively control and adjust the refrigeration system according to the physical thermal state of the host system in a large-scale computing center, realize real-time, refined and intelligent control of the energy consumption of the refrigeration system, and effectively meet the requirements of energy conservation and emission reduction.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows: A method for optimizing the energy consumption of a refrigeration system for a computing center, comprising the following steps: S101, configure the regulation environment and its regulation factors required to be enabled by the refrigeration system; S102. Measure the thermal state of the host system of the measurement and calculation center and the environmental information of the regulation environment. S103. Based on the thermal state of the host system and the environmental information of the regulation environment, use a pre-trained refrigeration control decision model to decide on the control actions for the regulation factors of the refrigeration system. Before step S103, it also includes training the refrigeration control decision model: S401. Construct a training data sample composed of the thermal state of the host system of the computing center, the environmental information of the regulation environment, the control actions of the regulation factors, and the refrigeration efficiency. The environmental information of the regulation environment includes some or all of the environmental temperature and environmental humidity of the regulation environment. S402. Train the refrigeration control decision model based on the Deep Deterministic Policy Gradient algorithm DDPG. Use the thermal state of the host system of the computing center and the environmental information of the regulation environment as the "state" of the refrigeration control decision model, use the regulation factors of the refrigeration system as the "actions" of the refrigeration control decision model, and the action regulation range is the regulation range corresponding to the regulation factors. Use the refrigeration efficiency as the "reward" of the refrigeration control decision model. Train the refrigeration control decision model through the training data sample to establish the mapping relationship between the thermal state of the host system, the environmental information of the regulation environment, and the control actions of the regulation factors of the refrigeration system.

[0007] Optionally, step S101 includes: S201. According to the preset default regulation environment, configure the required enabled regulation environment. S202. For each enabled regulation environment, configure the subordinate regulation factors respectively. Each configured regulation factor includes the regulation factor name and the physical device parameter identifier corresponding to the regulation factor. The regulation factors subordinate to the host system include some or all of the terminal supply air temperature, terminal supply air volume, terminal supply water temperature, and terminal supply water volume. The regulation factors subordinate to the refrigeration system include some or all of the chiller power, pump power, water supply volume, fan speed, and water pump flow rate. The regulation factors subordinate to the computer room environment include some or all of the computer room air conditioner temperature and computer room air conditioner humidity. The regulation factors subordinate to the outdoor environment include some or all of the outdoor air ventilation volume and outdoor ventilation duration. S203. Determine the configuration regulation range for each regulation factor. The regulation range is a triple < u, f, s >>, where u represents the regulation upper limit, f represents the regulation lower limit, s represents the adjustable amplitude each time.

[0008] Optionally, the default regulation environment preset in step S201 is represented by a quadruple: <S, C, H, T>, where S represents the host system; C represents the refrigeration system; H represents the computer room environment; T represents the outdoor environment. The configured regulation environment to be enabled includes, for each regulation environment in the preset default regulation environment, using key-value pairs composed of the regulation environment and whether it is included in the regulation strategy to configure the regulation environment to be enabled. The value of "whether it is included in the regulation strategy" being "1" indicates inclusion in the regulation strategy, and a value of "0" indicates non-inclusion in the regulation strategy.

[0009] Optionally, measuring the thermal state of the host system in the computing center in step S102 includes: S301, using the Baseboard Management Controller (BMC) installed and configured on the multi-source heterogeneous computing devices of the host system to collect the temperature data of each temperature sensor on the computing devices, and performing out-of-band collection of the temperature data of the computing devices based on the IPMI protocol; S302, using a preset aggregation algorithm on the set of temperature data of the computing devices to determine whether the computing device has a temperature anomaly; S303, adding up the number of computing devices with temperature anomalies in each preset area on the host system to obtain the total number of computing devices with temperature anomalies in the host system of the computing center, as the thermal state of the host system of the obtained computing center.

[0010] Optionally, in step S302, using a preset aggregation algorithm on the set of temperature data of the computing devices to determine whether the computing device has a temperature anomaly includes: using a preset statistical method to calculate the aggregated temperature value of the computing device for the set of temperature data of the computing device, and then determining whether the computing device has a temperature anomaly based on whether the aggregated temperature value exceeds a preset threshold.

[0011] Optionally, the calculation function expression of the refrigeration efficiency is: , In the above formula, represents the refrigeration efficiency, is a penalty factor with a negative value, is the thermal state of the host system of the computing center, is the energy efficiency of the computing center, and the energy efficiency of the computing center is the ratio of all energy consumption of the computing center to the energy consumption of the host system.

[0012] In addition, the present invention also provides a refrigeration system energy consumption optimization system for a computing center, including a microprocessor and a memory connected to each other, and the microprocessor is programmed or configured to execute the refrigeration system energy consumption optimization method for the computing center.

[0013] In addition, the present invention further provides a computer-readable storage medium storing a computer program or instruction, which is programmed or configured to execute the method for optimizing the energy consumption of the refrigeration system for a computing center through a processor.

[0014] In addition, the present invention further provides a computer program product including a computer program or instruction, which is programmed or configured to execute the method for optimizing the energy consumption of the refrigeration system for a computing center through a processor.

[0015] Compared with the prior art, the present invention mainly has the following advantages: The method for optimizing the energy consumption of the refrigeration system for a computing center according to the present invention includes configuring the regulation environment and its regulation factors required for enabling the refrigeration system; measuring the thermal state of the host system in the computing center and the environmental information of the regulation environment; using a pre-trained refrigeration control decision model to make decisions on the control actions of the regulation factors of the refrigeration system based on the thermal state of the host system and the environmental information of the regulation environment. The present invention makes decisions on the control actions of the regulation factors of the refrigeration system through the thermal state of the host system and the environmental information of the regulation environment, so as to be able to achieve adaptive control and adjustment of the refrigeration system according to the physical thermal state of the host system in a large-scale computing center, be able to minimize the energy consumption of the refrigeration system on the premise of ensuring the normal operation of the host system, realize real-time, refined, and intelligent regulation of the energy consumption of the refrigeration system, and effectively meet the requirements of energy conservation and emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the basic process of the method in Embodiment 1 of the present invention.

[0017] Figure 2 It is a schematic diagram of the basic principle of the method in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] Embodiment 1: As Figure 1 shown, the method for optimizing the energy consumption of the refrigeration system for a computing center in this embodiment includes the following steps: S101, configure the regulation environment and its regulation factors required for enabling the refrigeration system; S102, measure the thermal state of the host system in the computing center and the environmental information of the regulation environment; S103, use a pre-trained refrigeration control decision model to make decisions on the control actions of the regulation factors of the refrigeration system based on the thermal state of the host system and the environmental information of the regulation environment.

[0019] In this embodiment, the operation and maintenance management personnel of the large-scale computing center perform the initialization of step S101 for the energy consumption optimization task through template-based parameter configuration. Specifically, step S101 in this embodiment includes: S201, configure the required enabled control environment according to the preset default control environment; S202, configure the subordinate control factors for the enabled control environment respectively, and each configured control factor includes a control factor name and a physical device parameter identifier corresponding to the control factor. For example, the configuration content of the terminal air conditioner temperature is: Terminal air conditioner temperature: end_ac_temp; As Figure 2 shown, for each environmental element incorporated into the control strategy, the operation and maintenance management personnel can further configure a specific set of control factors. For example, the control factors subordinate to the host system include some or all of the terminal supply air temperature, terminal supply air volume, terminal supply water temperature, and terminal supply water volume. The control factors subordinate to the refrigeration system include some or all of the chiller power, pump power, water supply volume, fan speed, and water pump flow rate. The control factors subordinate to the computer room environment include some or all of the computer room air conditioner temperature and computer room air conditioner humidity. The control factors subordinate to the outdoor environment include some or all of the outdoor air ventilation volume and outdoor ventilation duration. Among them, the control factors subordinate to the host system are related to the refrigeration mechanism of the host system. For example, as an optional implementation method, in this embodiment, the refrigeration mechanism of the host system includes a host air-cooled unit and a host water-cooled unit. The host air-cooled unit introduces low-temperature natural wind from the outdoor environment through a fan at the terminal and sends it into each host system in the computer room through a pipeline. The host water-cooled unit sends the chilled water after being refrigerated by the refrigeration mechanism into each host system in the computer room through a pump at the terminal. Therefore, the terminal supply air temperature, terminal supply air volume, terminal supply water temperature, and terminal supply water volume can be used as the control factors subordinate to the host system. The control factors subordinate to the refrigeration system are related to the refrigeration system of the computer room. The refrigeration system of the computer room generally uses a dual-medium circulation loop, and heat exchange is carried out between the two dual-medium circulation loops through a heat exchanger. The first medium circulation loop is the refrigerant, which is circulated by a pump and refrigerated by a chiller. The second medium circulation loop is water, which is circulated by a water pump and exchanges heat with the computer room through a fan. Therefore, the chiller power, pump power, water supply volume, fan speed, and water pump flow rate can be used as the control factors subordinate to the refrigeration system. The control factors subordinate to the computer room environment are related to the air conditioning system of the computer room environment. Therefore, the computer room air conditioner temperature and computer room air conditioner humidity can be used as control factors. The control factors subordinate to the outdoor environment are related to the ventilation method of the computer room. Therefore, the outdoor air ventilation volume and outdoor ventilation duration can be used as control factors; S203, determine the configured control range for each control factor. The control range is a triple < u, f, s >, where u represents the control upper limit, f represents the control lower limit, s represents the adjustable amplitude each time.

[0020] In step S201 of this embodiment, the preset default control environment is represented by a quadruple: <S, C, H, T>, where S represents the host system; C represents the refrigeration system; H represents the computer room environment; T represents the outdoor environment. The configured control environment to be enabled includes, for each control environment in the preset default control environment, configuring the control environment to be enabled using key-value pairs composed of the control environment and whether it is included in the control strategy. The value of whether it is included in the control strategy being "1" means it is included in the control strategy, and the value being "0" means it is not included in the control strategy. For example: { S:1 C:1 H:1 T:0 } It means that the outdoor environment is configured not to be included in the control strategy, and the other three control environments are included in the control strategy.

[0021] In step S102 of this embodiment, measuring the thermal state of the host system of the computing center includes: S301, using the Baseboard Management Controller (BMC) installed and configured on the multi-source heterogeneous computing devices of the host system to collect the temperature data of each temperature sensor on the computing devices, and performing out-of-band collection of the temperature data of the computing devices based on the Intelligent Platform Management Interface (IPMI) protocol; S302, using a preset aggregation algorithm on the set of temperature data of the computing devices to determine whether the computing devices have abnormal temperatures; according to the composition structure and topological relationship of the host system, dividing the system into regions at the granularity of chassis or cabinets, etc., so as to determine each preset region on the host system. And data aggregation modules are respectively set in each region, and this module is responsible for summarizing and aggregating the original temperature data of all computing devices in the region. The aggregation value is the number of computing devices with abnormal temperatures in the region. The aggregation module reports the aggregation value to the previous control region to improve data transmission efficiency; S303, adding up the number of computing devices with abnormal temperatures in each preset region on the host system to obtain the total number of computing devices with abnormal temperatures in the host system of the computing center, which is used as the thermal state of the host system of the obtained computing center.

[0022] In step S302 of this embodiment, the preset aggregation algorithm is designed in a plug-in manner, which enables operation and maintenance management personnel to conveniently configure the required aggregation algorithm. The purpose of the aggregation algorithm is to aggregate and map multiple temperature data of the computing device into whether the computing device has a temperature anomaly, and the specific mapping method can be selected as needed. For example, as an optional implementation manner, in step S302 of this embodiment, determining whether the computing device has a temperature anomaly for the temperature data set of the computing device by using the preset aggregation algorithm includes: calculating the aggregated temperature value of the computing device by using a preset statistical method (such as taking the average value, median value, maximum value, minimum value, etc. in the set) for the temperature data set of the computing device, and then determining whether the computing device has a temperature anomaly according to whether the aggregated temperature value exceeds a preset threshold.

[0023] The cold control decision model can adopt the required machine learning model as needed, and establish the mapping relationship between the thermal state of the host system, the environmental information of the regulation environment, and the control actions of the regulation factors of the refrigeration system through training. Before step S103 of this embodiment, training the refrigeration control decision model is also included: S401, constructing a training data sample composed of the thermal state of the host system in the computing center, the environmental information of the regulation environment, the control actions of the regulation factors, and the refrigeration efficiency, where the environmental information of the regulation environment includes some or all of the environmental temperature and environmental humidity of the regulation environment; S402, train the refrigeration control decision-making model based on the Deep Deterministic Policy Gradient (DDPG) algorithm, and use the thermal state of the host system in the computing center and the environmental information of the regulation environment as the "state" of the refrigeration control decision-making model, use the regulation factors of the refrigeration system as the "action" of the refrigeration control decision-making model, and the action regulation range is the regulation range corresponding to the regulation factor. Use the refrigeration efficiency as the "reward" of the refrigeration control decision-making model. Train the refrigeration control decision-making model through training data samples to establish the mapping relationship between the thermal state of the host system, the environmental information of the regulation environment, and the control actions of the regulation factors of the refrigeration system. When training the refrigeration control decision-making model based on the DDPG algorithm, the input of the model requires three parts of information, namely "state", "action" and "reward". Among them, the "state" is the environmental information of the regulation environment configured in step S101 and the thermal state of the host system; the "action" is the regulation factor of the refrigeration system configured in step S101, and the action regulation range is the regulation range corresponding to the regulation factor; the "reward" is the evaluation value of the refrigeration efficiency. The trained decision-making model can directly output the optimal action to be taken next according to the input state, realizing the adaptive regulation of the refrigeration equipment. In this embodiment, the problem of optimizing the energy consumption of the refrigeration system is modeled as a Markov decision process to train and complete the mapping of "state" and "action" in combination with "reward", so as to adjust the operating parameters of the refrigeration system according to the current thermal state of the host system, and can minimize the energy consumption of the refrigeration system on the premise of ensuring the normal operation of the host system.

[0024] In this embodiment, the refrigeration efficiency is evaluated by customizing based on the thermal state of the host system and the system energy consumption, and the control strategy is adjusted through the reinforcement learning model. As an alternative implementation, the goal of optimizing the energy consumption of the refrigeration system is to minimize the energy consumption of the refrigeration system on the premise of meeting the normal operation of the host system. The calculation function expression of the refrigeration efficiency is as follows: , In the above formula, represents the refrigeration efficiency, is a penalty factor with a negative value, is the thermal state of the host system in the computing center, (Power Usage Effectiveness) is the energy efficiency of the computing center. The energy efficiency of the computing center is the ratio of all energy consumption of the computing center to the energy consumption of the host system. The above formula shows that when there are computing devices with abnormal temperatures in the host system, the refrigeration efficiency value is the product of the number of abnormal components and the penalty factor. When all computing devices in the host system are within the normal temperature range, the refrigeration efficiency value is the reciprocal of PUE, that is, the lower the PUE value, the higher the refrigeration efficiency.

[0025] In summary, the method for optimizing the energy consumption of the refrigeration system for the computing center in this embodiment is based on the configured refrigeration energy consumption optimization task generation framework, which improves the applicability and flexibility of the method in this embodiment and can be widely applied to different large-scale computing centers. The method in this embodiment is based on the host system thermal state measurement method of software-hardware cooperation and regional data aggregation, which improves the acquisition and processing efficiency of large-scale heat source data, enabling the control program to quickly respond to changes in the host system state. The method in this embodiment is a multi-refrigeration device collaborative control method based on reinforcement learning, which improves the self-adaptability and intelligence level of the refrigeration energy consumption optimization process and can continuously and real-time automatically generate the optimal refrigeration control strategy according to the host system thermal state and the operation of the refrigeration equipment. This embodiment makes decisions on the control actions of the control factors of the refrigeration system through the host system thermal state and the environmental information of the control environment, so as to realize the adaptive control and adjustment of the refrigeration system according to the physical thermal state of the host system in the large-scale computing center, ensure the normal operation of the host system, minimize the energy consumption of the refrigeration system, and realize the real-time, refined and intelligent control of the energy consumption of the refrigeration system, effectively meeting the requirements of energy conservation and emission reduction.

[0026] Embodiment 2: This embodiment is basically the same as Embodiment 1, and the main difference is that the method of configuring the required enabled control environment is different. In this embodiment, the configured required enabled control environment includes, for each control environment in the preset default control environment, configuring the required enabled control environment by using a fixed order of whether to be included in the control strategy. The value of whether to be included in the control strategy being "1" means being included in the control strategy, and the value being "0" means not being included in the control strategy. For example: {1,1,1,0} It means that the configuration is such that the outdoor environment is not included in the control strategy, and the other three control environments are included in the control strategy.

[0027] Embodiment 3: This embodiment is basically the same as Embodiment 1, and the main difference is that in step S302 of this embodiment, the preset aggregation algorithm uses a machine learning model. The machine learning model is trained to establish the relationship between multiple temperature data of the computing device and whether the computing device has a temperature anomaly. When in use, the multiple temperature data of the computing device are input into the machine learning model to obtain the prediction result of whether the computing device has a temperature anomaly. Through the above method, the relationship mapping between the multiple temperature data of the computing device and whether the computing device has a temperature anomaly can also be realized.

[0028] Embodiment 4: This embodiment is basically the same as Embodiment 1, and the main difference is that: in this embodiment, the cooling control decision model is a machine learning model trained by a conventional supervised training method. Through supervised training, a mapping relationship is established among the thermal state of the host system, the environmental information of the regulation environment, and the control actions of the regulation factors of the refrigeration system. Similarly, a mapping among the thermal state of the host system, the environmental information of the regulation environment, and the control actions of the regulation factors of the refrigeration system can also be achieved.

[0029] In addition, this embodiment also provides a refrigeration system energy consumption optimization system for a computing center, including a microprocessor and a memory connected to each other. The microprocessor is programmed or configured to execute the refrigeration system energy consumption optimization method for the computing center. In addition, this embodiment also provides a computer-readable storage medium, in which a computer program or instruction is stored. The computer program or instruction is programmed or configured to execute the refrigeration system energy consumption optimization method for the computing center through a processor. In addition, this embodiment also provides a computer program product, including a computer program or instruction, and the computer program or instruction is programmed or configured to execute the refrigeration system energy consumption optimization method for the computing center through a processor.

[0030] Those skilled in the art should understand that the technical solution of the present invention can be provided in the form of a method, a system, or a computer program product. Therefore, the present invention can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can adopt the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes. The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 a device for realizing the functions specified in one block or multiple blocks. These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device realizes the functions in Figure 1 one flow or multiple flows and / or blocksFigure 1 The functions specified in one or more boxes. These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide for implementing in the process Figure 1 One process or multiple processes and / or boxes Figure 1 The steps of the functions specified in one or more boxes.

[0031] The above is only the preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements should also be regarded as within the protection scope of the present invention.

Claims

1. A method for optimizing energy consumption of a refrigeration system in a computing center, characterized in that: The steps include: S101, configuring the control environment and control factors required to be enabled for the refrigeration system; S102, measuring the thermal status of the host system of the computing center and the environmental information of the control environment; S103, using a pre-trained refrigeration control decision model to decide a control action on a control factor of the refrigeration system based on the thermal state of the host system and environmental information of the control environment; Before step S103, the process also includes training the refrigeration control decision model: S401, constructing a training data sample consisting of the thermal state of the host system of the computing center and the environmental information of the control environment, the control action of the control factor, and the cooling efficiency, wherein the environmental information of the control environment includes part or all of the ambient temperature and ambient humidity of the control environment; S402, training the refrigeration control decision model based on the deep deterministic policy gradient algorithm DDPG, and taking the thermal state of the host system of the computing center and the environmental information of the control environment as the "state" of the refrigeration control decision model, taking the control factor of the refrigeration system as the "action" of the refrigeration control decision model, and the action control amplitude is the control range corresponding to the control factor, taking the refrigeration benefit as the "reward" of the refrigeration control decision model, and training the refrigeration control decision model through training data samples to establish a mapping relationship between the thermal state of the host system and the environmental information of the control environment, and the control action of the control factor of the refrigeration system.

2. The method for optimizing energy consumption of a refrigeration system for a computing center according to claim 1, characterized in that: Step S101 includes: S201, configuring a control environment to be enabled according to a preset default control environment; S202, respectively configuring subordinate control factors for the enabled control environment, and each configured control factor includes a control factor name and a physical device parameter identifier corresponding to the control factor; the control factors subordinate to the host system include part or all of the terminal air supply temperature, terminal air supply volume, terminal water supply temperature, and terminal water supply volume; the control factors subordinate to the refrigeration system include part or all of the refrigerator power, pump power, water supply volume, fan speed, and water pump flow rate; the control factors subordinate to the computer room environment include part or all of the computer room air conditioning temperature and the computer room air conditioning humidity; the control factors subordinate to the outdoor environment include part or all of the outdoor air ventilation volume and the outdoor ventilation time; S203, determining a configuration control range for each control factor, where the control range is a triplet < u, f, s >, where u represents the upper limit of regulation, f Indicates the lower limit of regulation, s Indicates the adjustable range each time.

3. The method for optimizing energy consumption of a refrigeration system for a computing center according to claim 2, characterized in that: The default control environment preset in step S201 is represented by a four-tuple:<S, C, H, T> , where S represents the host system; C represents the refrigeration system; H represents the computer room environment; and T represents the outdoor environment. The configuration of the control environment to be enabled includes configuring the control environment to be enabled for each control environment in the preset default control environment using a key-value pair consisting of a control environment and whether it is included in the control strategy. The value of whether it is included in the control strategy is "1" indicating that it is included in the control strategy, and the value of "0" indicates that it is not included in the control strategy.

4. The method for optimizing energy consumption of a refrigeration system for a computing center according to claim 1, characterized in that: Measuring the thermal state of the host system of the computing center in step S102 includes: S301, using a motherboard management controller BMC installed and configured on a multi-source heterogeneous computing device of a host system to collect temperature data of each temperature sensor on the computing device, and performing out-of-band collection of the temperature data of the computing device based on an IPMI protocol; S302, using a preset aggregation algorithm to determine whether the temperature of the computing device is abnormal. S303, adding the number of computing devices with abnormal temperature in each area preset on the host system to obtain the total number of computing devices with abnormal temperature of the host system as the obtained thermal state of the host system of the computing center.

5. The method for optimizing energy consumption of a refrigeration system for a computing center according to claim 4, characterized in that: In step S302, the temperature data set of the computing device is subjected to a preset aggregation algorithm to determine whether the computing device has a temperature abnormality, including: using a preset statistical method to calculate the aggregated temperature value of the computing device for the temperature data set of the computing device, and then determining whether the computing device has a temperature abnormality based on whether the aggregated temperature value exceeds a preset threshold.

6. The method for optimizing energy consumption of a refrigeration system for a computing center according to claim 1, characterized in that: The calculation function expression of the refrigeration benefit is: ， In the above formula, Represents the cooling efficiency, is a penalty factor with a negative value, is the thermal status of the host system in the computing center, The energy efficiency of the computing center is the ratio of all energy consumption of the computing center to the energy consumption of the host system.

7. A refrigeration system energy consumption optimization system for a computing center, comprising a microprocessor and a memory connected to each other, characterized in that: The microprocessor is programmed or configured to execute the method for optimizing energy consumption of a refrigeration system for a computing center as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program or instruction stored therein, characterized in that: The computer program or instruction is programmed or configured to execute the method for optimizing the energy consumption of a refrigeration system for a computing center as claimed in any one of claims 1 to 6 through a processor.

9. A computer program product comprising a computer program or instructions, characterized in that The computer program or instruction is programmed or configured to execute the method for optimizing the energy consumption of a refrigeration system for a computing center as claimed in any one of claims 1 to 6 through a processor.

Images