Energy-saving optimization method for liquid cooling system of data center

Abstract

The invention provides an energy-saving optimization method for a liquid cooling system of a data center, and relates to the technical field of liquid cooling energy-saving regulation and control. The method comprises the following steps: running a digital twinborn model, outputting a first regulation and control strategy and a corresponding PUE value, and recording the first regulation and control strategy and the corresponding PUE value as PUE1; inputting virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain a second regulation and control strategy and a corresponding PUE value, and recording the second regulation and control strategy and the corresponding PUE value as PUF2; and comparing the PUF1 with the PUF2 to obtain a final regulation and control strategy, and controlling the liquid cooling system of the data center according to the final regulation and control strategy. The two models are combined to realize accurate regulation and control, when the data acquisition sensor is abnormal, the digital twinborn model and the deep reinforcement learning model can still operate continuously, continuous regulation and control are realized, the problem of regulation and control missing or lag is avoided, and energy waste is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of liquid cooling energy-saving regulation, and in particular, to an energy-saving optimization method for a liquid cooling system in a data center. Background Art

[0002] In recent years, computing power has become the core force driving the green development of the digital economy. At the same time, the development of computing power has also led to a continuous and substantial increase in the energy consumption of data centers. The cooling system is one of the main sources of energy consumption in data centers, usually accounting for 30%-50% of the total energy consumption. Therefore, energy-saving optimization of the cooling system is of great significance for reducing the energy consumption of data centers and improving energy utilization efficiency.

[0003] The control method of the existing liquid cooling system in data centers adjusts the liquid cooling system based on sensor data, mainly by comparing the monitored data with preset thresholds, and performing control adjustments once a deviation is found.

[0004] This method mainly relies on expert knowledge and historical experience, and the regulation is based on set fixed thresholds. Facing changes in the environment, it cannot make optimal adjustments in a timely and accurate manner, easily causing energy waste; moreover, this method relies on the real-time and accurate monitoring of sensors and the real-time transmission of data. Once a sensor fails or data transmission is delayed, the control system will lack input data, resulting in missing or lagging regulation, and then causing overheating or overcooling of IT equipment, which will also cause energy waste. Summary of the Invention

[0005] The problem to be solved by the present invention is that the control method of the existing liquid cooling system in data centers relies on sensors and preset thresholds for regulation, lacks flexible adaptability to changing environments, and will cause energy waste.

[0006] To solve the above problems, in a first aspect, the present invention provides an energy-saving optimization method for a liquid cooling system in a data center, including:

[0007] Running a digital twin model of the data center refrigeration system, outputting a first regulation strategy and the corresponding PUE value, denoted as PUE1;

[0008] Inputting the virtual operation data of the digital twin model into a trained deep reinforcement learning model to obtain a second regulation strategy and the corresponding PUE value, denoted as PUE2;

[0009] Comparing PUE1 and PUE2 to obtain a final regulation strategy, and controlling the data center liquid cooling system according to the final regulation strategy.

[0010] Optionally, before running the digital twin model of the data center refrigeration system and outputting a first regulation strategy and the corresponding PUE value, denoted as PUE1, it further includes:

[0011] Determine whether the data acquisition sensors in the data center cooling system are normal;

[0012] When the data acquisition sensors are normal, obtain the real-time operation data of the data center liquid cooling system;

[0013] Optimize the constructed digital twin model using the real-time operation data, and then run the digital twin model;

[0014] When the data acquisition sensors are abnormal, directly run the digital twin model.

[0015] Optionally, the construction process of the digital twin model includes:

[0016] Obtain the equipment parameters and historical operation data in the data center cooling system;

[0017] Construct and adjust the digital twin model according to the equipment parameters and the historical operation data.

[0018] Optionally, the energy-saving optimization method for the data center liquid cooling system further includes:

[0019] Preprocess the historical operation data.

[0020] Optionally, the training of the deep reinforcement learning model includes:

[0021] Input the virtual operation data into the deep reinforcement learning model, set the coolant flow rate as the action variable, and train the deep reinforcement learning model with the goal of obtaining the minimum PUE value to obtain the trained deep reinforcement learning model.

[0022] Optionally, comparing PUE1 and PUE2 to obtain the final regulation strategy, and controlling the data center liquid cooling system according to the final regulation strategy includes:

[0023] If PUE1 is greater than PUE2, control the data center cooling system according to the second regulation strategy, and optimize the digital twin simulation model based on the real-time operation data of the regulated data center cooling system;

[0024] If PUE1 is less than or equal to PUE2, control the data center cooling system according to the first regulation strategy, and optimize the deep reinforcement learning model based on the virtual operation data of the regulated digital twin simulation model.

[0025] In a second aspect, the present invention also provides an energy-saving optimization system for a data center liquid cooling system, including:

[0026] A data twin simulation module, which is used to run the digital twin model of the data center cooling system, output the first regulation strategy and the corresponding PUE value, denoted as PUE1;

[0027] A deep reinforcement learning module, which is used to input the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain the second regulation strategy and the corresponding PUE value, denoted as PUE2;

[0028] A cooling system regulation module, which is used to compare PUE1 and PUE2 to obtain the final regulation strategy, and control the data center liquid cooling system according to the final regulation strategy.

[0029] Optionally, the energy-saving optimization system of the data center liquid cooling system further includes: a sensor monitoring module, a data collection module, a data preprocessing module, a digital twin model construction module, and a digital twin model optimization module;

[0030] The sensor monitoring module is used to judge whether the data acquisition sensors in the data center cooling system are normal;

[0031] The data collection module is used to obtain the equipment parameters and historical operation data in the data center cooling system;

[0032] The data preprocessing module is used to preprocess the historical operation data;

[0033] The digital twin model construction module is used to construct and adjust the digital twin model according to the equipment parameters and the historical operation data;

[0034] The data collection module is further used to obtain the real-time operation data of the data center liquid cooling system when the data acquisition sensors are normal;

[0035] The digital twin model optimization module is used to continuously optimize the constructed digital twin model by using the real-time operation data to obtain an optimized digital twin model.

[0036] In a third aspect, the present invention provides an electronic device, including a memory and a processor;

[0037] The memory is used to store a computer program;

[0038] The processor is used to implement the energy-saving optimization method of the data center liquid cooling system as described in the first aspect when executing the computer program.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the energy-saving optimization method of the data center liquid cooling system as described in the first aspect is implemented.

[0040] The present invention provides an energy-saving optimization method for a data center liquid cooling system. Compared with the prior art, it has the following beneficial effects:

[0041] When the data acquisition sensor is abnormal, the digital twin model can still continue to run, output the first regulation strategy and the corresponding PUE value, denoted as PUE1; input the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain the second regulation strategy and the corresponding PUE value, denoted as PUE2; even if the data acquisition sensor is damaged, the digital twin model can still continue to run and continue to output virtual operation data, so that the deep reinforcement learning model will not interrupt running, and thus PUE1 and PUE2 can still be obtained. Compare PUE1 and PUE2 to obtain the final regulation strategy, and control the data center liquid cooling system according to the final regulation strategy to achieve continuous and uninterrupted regulation, avoid the problems of regulation missing or lagging, and reduce energy waste; with the powerful self-learning ability of the deep reinforcement learning model, it can generate a more advantageous second regulation strategy and PUE2 value; combine the digital twin model with the deep reinforcement learning model, give full play to the advantages of both, and select the optimal strategy to achieve more accurate regulation, thereby effectively reducing energy waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 It is a schematic flowchart of an energy-saving optimization method for a data center liquid cooling system provided by an embodiment of the present invention;

[0044] Figure 2 It is a schematic overall flowchart of an energy-saving optimization method for a data center liquid cooling system provided by an embodiment of the present invention;

[0045] Figure 3 It is a schematic structural diagram of an energy-saving optimization system for a data center liquid cooling system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without making creative efforts belong to the scope of protection of this application.

[0047] To better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the accompanying drawings of the specification and specific implementation manners.

[0048] As Figure 1 shown, a method for energy-saving optimization of a data center liquid cooling system provided by an embodiment of this application includes:

[0049] S1: Run the digital twin model of the data center refrigeration system, and output the first regulation strategy and the corresponding PUE value (Power Usage Effectiveness), denoted as PUE1;

[0050] S2: Input the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain the second regulation strategy and the corresponding PUE value, denoted as PUE2;

[0051] S3: Compare PUE1 and PUE2 to obtain the final regulation strategy, and control the data center liquid cooling system according to the final regulation strategy.

[0052] In this embodiment, through the digital twin model, the operation of the data center cooling system can be effectively simulated, and the possible operation of the data center cooling system in the future can also be pre-enacted. It is applicable to various operating environments and can effectively achieve the timely monitoring of the data center. When the data acquisition sensor is abnormal, the digital twin model can still continue to run and output the first regulation strategy and the corresponding PUE value, denoted as PUE1; input the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain the second regulation strategy and the corresponding PUE value, denoted as PUE2; even if the data acquisition sensor is damaged, the digital twin model can still continue to run and continue to output virtual operation data, so that the deep reinforcement learning model will not interrupt its operation. Therefore, PUE1 and PUE2 can still be obtained. Compare PUE1 and PUE2 to obtain the final regulation strategy, and control the data center liquid cooling system according to the final regulation strategy to achieve continuous and uninterrupted regulation, avoid the problems of regulation omission or lag, and reduce energy waste; with the powerful self-learning ability of the deep reinforcement learning model, a more advantageous second regulation strategy and PUE2 value can be generated; combine the digital twin model with the deep reinforcement learning model, give full play to the advantages of both, select the optimal strategy, and achieve more accurate regulation, thereby effectively reducing energy waste.

[0053] The following describes each step in detail.

[0054] 1. Data collection

[0055] Obtain the equipment parameters, historical operation data, and real-time operation data of the data center liquid cooling system.

[0056] Specifically, this part of the data is mainly widely collected from a large amount of historical data in the data center equipment operation logs, work logs, and meteorological information, including equipment model parameters, equipment operation parameters, corresponding environmental data, and energy consumption data. At the same time, use sensors to collect the real-time operation data of the data center operation, including environmental temperature, humidity, coolant temperature, coolant flow rate, and pipeline pressure value, etc.

[0057] 2. Data preprocessing

[0058] Preprocess the collected historical operation data, carefully check for outliers and missing values in the data, and use methods such as linear filling for correction and supplementation to ensure the correctness and integrity of the data, providing reliable data support for the subsequent training of the digital twin model.

[0059] 3. Digital twin model

[0060] Construct and adjust the digital twin model according to the equipment parameters and the historical operation data. By constructing the digital twin model, simulate the operation of the data center to achieve efficient monitoring, adjustment, and optimization of the real data center.

[0061] Specifically, compare the virtual operation data of the digital twin model with the continuously collected real-time operation data, and continuously adjust the parameters of the digital twin model to make the virtual operation data of the digital twin model coincide with the real-time operation data, so as to obtain a well-trained digital twin model.

[0062] According to equipment model parameters, equipment operation data, environmental data, etc., construct a digital twin model that can accurately reflect the operation status of the data center. Run simulation software (such as 6SigmaDC), and build a model by coupling CFD technology (Computational Fluid Dynamics) with physics-informed neural networks (PINNs) to simulate the flow and heat transfer process of the coolant in the refrigeration equipment, so as to realize real-time monitoring, precise adjustment and optimization of the data center refrigeration system. The collected real-time operation data (such as environmental temperature, humidity, etc.) can be transmitted to the corresponding position of the digital twin model to test and adjust the digital twin model.

[0063] Since the digital twin model simulates the operation status of the data center refrigeration system, when the virtual operation data is abnormal, the operation of the data center refrigeration system may also be abnormal or about to be abnormal. Therefore, the first control strategy can be executed to make the abnormal data center refrigeration system return to normal, and at the same time update and optimize the digital twin model to make the virtual operation data return to normal.

[0064] 3. Deep reinforcement learning model

[0065] Taking the reduction of the data center PUE value as the optimization goal, use the virtual operation data output by the digital twin model to train the model, and then predict the IT equipment energy consumption value of the data center based on the trained model and optimize the operation value of the refrigeration equipment to achieve energy-saving effects.

[0066] Input the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain the second control strategy and the corresponding PUE value, denoted as PUE2.

[0067] Specifically, input the virtual operation data into the deep reinforcement learning model (Deep Reinforcement Learning, abbreviated as DRL), set the coolant flow rate as the action variable, and take the minimization of the PUE value as the optimization goal to train the deep reinforcement learning model to obtain the trained deep reinforcement learning model.

[0068] A model is constructed using a deep reinforcement learning algorithm. Considering the actual application scenarios and data characteristics comprehensively, the model structure and the effective value ranges of various operating data are reasonably determined. Based on the virtual operating data within the digital twin model, such as environmental temperature, humidity, IT equipment energy consumption, coolant temperature, and pipe internal pressure values, etc., these data are used as state parameters and input into the deep reinforcement learning model, and the coolant flow rate is set as the action variable. During the training process of the deep reinforcement learning model, there are reward values and penalty values. For a certain state parameter, when modifying the action variable value, for the action variable value with a smaller PUE value, increase its reward value; for the action variable value with a larger PUE value, increase its penalty value; when the variation range of the PUE value is less than the threshold after multiple trainings, stop the training. By continuously adjusting the internal parameters of the model, an optimally trained model is finally obtained. This model can output a qualified coolant flow rate, and can also make the PUE value of the data center liquid cooling system reach the optimal, and output the optimal regulation strategy.

[0069] Based on the deep reinforcement learning model, through the processing and analysis of a large amount of virtual operating data, accurate prediction of the energy consumption of the data center liquid cooling system is realized, and the optimal strategy is output. Compared with the traditional adjustment method, this optimization adjustment method based on deep reinforcement learning can significantly improve the accuracy and forward-looking of regulation, and achieve the energy-saving goal.

[0070] 4. Refrigeration equipment operation control strategy

[0071] This part runs through the entire solution process, and different control strategies are adopted according to different situations, such as Figure 2 As shown, it is necessary to first judge whether the data acquisition sensors in the data center refrigeration system are normal.

[0072] (1) When both the sensor and data transmission are normal

[0073] When the data acquisition sensor is normal, obtain the real-time operating data; optimize the constructed digital twin model using the real-time operating data; then run the optimized digital twin model to generate the first regulation strategy and obtain the corresponding PUE value, denoted as PUE1. Then compare PUE1 and PUE2 to obtain the final regulation strategy, and control the data center liquid cooling system according to the final regulation strategy.

[0074] (2) When the sensor fails or the data transmission is abnormal

[0075] When the data acquisition sensor is abnormal, skip the optimization of the digital twin model and directly run the current digital twin model to generate the first regulation strategy and obtain the corresponding PUE value, denoted as PUE1. Then compare PUE1 and PUE2 to obtain the final regulation strategy, and control the data center liquid cooling system according to the final regulation strategy.

[0076] Specifically, the data acquisition sensor failure can be faults such as sensor damage, disconnection of the sensor transmission link, and power-off of the sensor; the abnormal transmission of the acquired data can be abnormal situations such as data transmission lag or data transmission loss. When such a situation occurs, although the digital twin model lacks real-time input data, the model can continue to run in simulation. At this time, the virtual operation data of the digital twin model is used to continue to support the operation of the deep learning model, so the first regulation strategy, the second regulation strategy, and PUE1 and PUE2 can still be obtained at this time. The regulation of the liquid cooling system of the data center will not terminate due to sensor abnormalities, and the regulation and monitoring can continue, which can minimize the losses caused by sensor failures or abnormal data transmission and ensure the normal and efficient operation of the data center.

[0077] In addition, when the digital twin model and the deep reinforcement learning model are running normally, at this time, the PUE values predicted by the two models are compared to decide the optimal adjustment plan. As Figure 2 shown, it specifically includes the following steps: Determine whether PUE1 is greater than PUE2. If PUE1 is greater than PUE2, then control the data center refrigeration system according to the second regulation strategy, and optimize the digital twin simulation model based on the real-time operation data of the regulated data center refrigeration system. If PUE1 is less than or equal to PUE2, then control the data center refrigeration system according to the first regulation strategy, and optimize the deep reinforcement learning model based on the virtual operation data of the regulated digital twin simulation model. As the data center refrigeration system operates for a long time, its operating state will gradually change. During the above regulation and continuous operation, the optimization of the digital twin model and the deep reinforcement learning model is still maintained, so that the two models can follow the operating state of the data center refrigeration system, making the implemented regulation strategy fit the data center refrigeration system and achieving better regulation effects.

[0078] In summary, compared with the prior art, the following beneficial effects are achieved:

[0079] 1. Application of digital twin technology: Based on digital twin technology, a simulation model is generated, and a professional simulation software is used to simulate the heat transfer process. By applying this technology to the energy-saving optimization system of the data center, the digital twin model can effectively simulate the operation of the data center, and can effectively achieve real-time monitoring, flexible adjustment, and in-depth optimization of the data center. When data transmission and acquisition are blocked, regulation is carried out based on the virtual operation data of the digital twin model, and a timely response is made, ensuring the operation stability of the data center, reducing unnecessary losses, and significantly improving the timeliness of regulation.

[0080] 2. Application of deep reinforcement learning algorithm: With the powerful self-learning ability of the deep reinforcement learning algorithm, the model is trained to make decisions. By applying this algorithm to the energy-saving optimization system of the data center, more advantageous regulation strategies can be generated.

[0081] 3. Technical integration and innovative application: Combine digital twin technology with deep reinforcement learning algorithms, give full play to the advantages of both to generate control strategies and conduct in-depth analysis and comparison, select the optimal strategy, and optimize the respective model parameters based on the analysis results to continuously improve the model performance. It is more flexible than traditional fixed control methods based on expert knowledge, can achieve more accurate control, and thus effectively reduce energy waste.

[0082] As Figure 3 shown, an energy-saving optimization system for a data center liquid cooling system provided by an embodiment of the present application includes:

[0083] A digital twin simulation module 600, configured to run a digital twin model of the data center refrigeration system and output a first control strategy and a corresponding PUE value, denoted as PUE1, when the data acquisition sensor is abnormal;

[0084] A deep reinforcement learning module 700, configured to input the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain a second control strategy and a corresponding PUE value, denoted as PUE2;

[0085] A refrigeration system control module 800, configured to compare PUE1 and PUE2 to obtain a final control strategy, and control the data center liquid cooling system according to the final control strategy.

[0086] In an alternative embodiment of the present application, the energy-saving optimization system for the data center liquid cooling system further includes: a sensor monitoring module 100, a data collection module 200, a data preprocessing module 300, a digital twin model construction module 400, and a digital twin model optimization module 500;

[0087] The sensor monitoring module 100 is configured to determine whether the data acquisition sensor in the data center refrigeration system is normal;

[0088] The data collection module 200 is configured to obtain the equipment parameters and historical operation data in the data center refrigeration system;

[0089] The data preprocessing module 300 is configured to preprocess the historical operation data;

[0090] The digital twin model construction module 400 is configured to construct and adjust a digital twin model according to the equipment parameters and the historical operation data;

[0091] The data collection module 200 is further configured to obtain the real-time operation data of the data center liquid cooling system when the data acquisition sensor is normal;

[0092] The data twin model optimization module 500 is configured to continuously optimize the constructed digital twin model by using real-time operation data, so as to obtain an optimized digital twin model.

[0093] In this embodiment, the beneficial effects of the data center liquid cooling system energy-saving optimization system are similar to those of the above-mentioned data center liquid cooling system energy-saving optimization method, and will not be elaborated here.

[0094] An electronic device provided by an embodiment of the present application includes a memory and a processor; the memory is used to store a computer program; the processor is configured to implement the above-mentioned data center liquid cooling system energy-saving optimization method when executing the computer program.

[0095] A computer-readable storage medium provided by an embodiment of the present application has a computer program stored thereon, and when the computer program is executed by a processor, the above-mentioned data center liquid cooling system energy-saving optimization method is implemented.

[0096] In this embodiment, the beneficial effects of the electronic device and the computer-readable storage medium are similar to those of the above-mentioned data center liquid cooling system energy-saving optimization method, and will not be elaborated here.

[0097] Now, an electronic device that can be used as a server or a client of the present application will be described. It is an example of a hardware device that can be applied to various aspects of the present application. The electronic device is intended to represent various forms of digital electronic computer devices, such as, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0098] The electronic device includes a computing unit, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) or a computer program loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The computing unit, the ROM, and the RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0099] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc. In this application, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. One can select some or all of the units according to actual needs to achieve the purpose of the solution of the embodiments of this application. In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0100] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0101] The above embodiments are only used to illustrate the technical solutions of this application, rather than to limit them; although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A data center liquid cooling system energy-saving optimization method, characterized in that: include: Run the digital twin model of the data center cooling system and output the first control strategy and the corresponding PUE value, recorded as PUE1; Inputting the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain a second control strategy and a corresponding PUE value, recorded as PUE2; PUE1 and PUE2 are compared to obtain a final control strategy, and the data center liquid cooling system is controlled according to the final control strategy.

2. The data center liquid cooling system energy saving optimization method according to claim 1, characterized in that: The digital twin model of the data center cooling system is operated, and before outputting the first control strategy and the corresponding PUE value, denoted as PUE1, it also includes: Determine whether the data acquisition sensors in the data center cooling system are normal; When the data acquisition sensor is normal, real-time operation data of the data center liquid cooling system is obtained; Utilizing real-time operation data to optimize the constructed digital twin model, and then running the digital twin model; When the data acquisition sensor is abnormal, the digital twin model is directly run.

3. The data center liquid cooling system energy saving optimization method according to claim 1, characterized in that: The construction process of the digital twin model includes: Obtain equipment parameters and historical operating data in the data center cooling system; A digital twin model is constructed and adjusted according to the equipment parameters and the historical operation data.

4. The data center liquid cooling system energy saving optimization method according to claim 3, characterized in that: After obtaining the equipment parameters and historical operation data in the data center refrigeration system, the method further includes: The historical operation data is preprocessed.

5. The data center liquid cooling system energy saving optimization method according to claim 1, characterized in that: The training of the deep reinforcement learning model includes: The virtual operation data is input into the deep reinforcement learning model, and the coolant flow rate is set as the action variable to obtain the minimum PUE value as the optimization target, and the deep reinforcement learning model is trained to obtain the trained deep reinforcement learning model.

6. The data center liquid cooling system energy saving optimization method according to claim 1, characterized in that: The comparing PUE1 and PUE2 to obtain a final control strategy, and controlling the data center liquid cooling system according to the final control strategy includes: If PUE1 is greater than PUE2, the data center cooling system is controlled according to the second control strategy, and the digital twin simulation model is optimized according to the real-time operation data of the data center cooling system after control; If PUE1 is less than or equal to PUE2, the data center refrigeration system is controlled according to the first control strategy, and the deep reinforcement learning model is optimized based on the virtual operation data of the digital twin simulation model after control.

7. A data center liquid cooling system energy saving optimization system, characterized in that: include: The data twin simulation module is used to run the digital twin model of the data center cooling system and output the first control strategy and the corresponding PUE value, denoted as PUE1; A deep reinforcement learning module, used for inputting the virtual operation data of the digital twin model into the trained deep reinforcement learning model to obtain a second control strategy and a corresponding PUE value, recorded as PUE2; The cooling system control module is used to compare PUE1 and PUE2 to obtain a final control strategy, and control the data center liquid cooling system according to the final control strategy.

8. The data center liquid cooling system energy saving optimization system according to claim 7, characterized in that: It also includes: sensor monitoring module, data collection module, data preprocessing module, digital twin model building module and data twin model optimization module; The sensor monitoring module is used to determine whether the data acquisition sensor in the data center refrigeration system is normal; The data collection module is used to obtain equipment parameters and historical operation data in the data center refrigeration system; The data preprocessing module is used to preprocess the historical operation data; The digital twin model construction module is used to construct and adjust the digital twin model according to the device parameters and the historical operation data; The data collection module is also used to obtain real-time operation data of the data center liquid cooling system when the data acquisition sensor is normal; The data twin model optimization module is used to continuously optimize the constructed digital twin model using real-time operation data to obtain an optimized digital twin model.

9. An electronic device, characterized in that: including memory and processor; The memory is used to store computer programs; The processor is used to implement the data center liquid cooling system energy-saving optimization method as described in any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the energy-saving optimization method for a data center liquid cooling system as described in any one of claims 1 to 6 is implemented.

Images