Data center energy-saving cooling control method and system

Abstract

The invention provides an energy-saving cooling control method and system for a data center. The energy-saving cooling control method comprises the steps that an initial temperature set value and a thermal safety temperature target of an air cooling device are obtained; predicting the power of the air cooling device through a pre-trained data center time sequence model based on the initial temperature set value; an ideal temperature set value corresponding to the minimum power value of the air cooling device is determined through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; and the ideal temperature set value is transmitted to a smooth buffer, a control instruction is sent to a controller of the air cooling device through the smooth buffer, and the power of the air cooling device is adjusted. The problem that an existing data center is large in cooling energy consumption is solved.

Description

Technical Field

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

[0002] The growing demand for artificial intelligence and cloud computing has led to skyrocketing energy consumption in data centers (DCs). Current industry practices involve human operators, and many data-driven approaches have been proposed. However, human intervention often leads to unnecessary energy consumption, especially in the face of fluctuating server loads, while existing data-driven approaches have difficulty maintaining thermal safety in practice.

[0003] The use of DC power is expected to increase by two or even four times by 2030. The substantial growth in energy demand not only poses challenges to environmental sustainability, but also increases the operating costs of commercial DC suppliers. Currently, the most commonly used control setting for air cooling units (ACUs) is their inlet temperature set point. To achieve a given set point, the ACU utilizes proportional-integral-derivative (PID) control. To select the set point, current industry practice relies on human operators to select a value based on experience, while many data-driven methods have been proposed that require learning how to calculate the temperature set point under dynamic load from historical data center trajectories. However, none of the existing technologies can achieve the goal of reducing data center cooling energy within the thermal safety range. Since the server load is unknown to the operator, human intervention cannot provide the optimal cooling energy. The mismatch between cooling supply and server load leads to unnecessary cooling energy consumption. Existing data-driven methods, although load-sensitive, have difficulty meeting thermal safety constraints because they ignore the dynamic characteristics of the air cooling unit PID controller. They take the minimization of cooling energy under thermal safety constraints as the optimization goal. This setting incentivizes the ACU to operate at the boundary of the constraint. As a result, cooling interruptions often occur. The set point is higher than the actual inlet temperature, so the PID controller thinks it has exceeded the set target and the cold air is delivered at a reduced or zero rate. This phenomenon does not meet the direct cooling standards set by the Society of Refrigerating and Air-Conditioning Engineers. When this happens, the temperature throughout the data center rises rapidly due to the heat generated by the servers being removed at a lower rate. Even if the cold air begins to be delivered at a faster rate again, it takes longer to undo the increase, which makes it impossible for previous data-driven methods to generate a set point to suppress the temperature rise in time, resulting in difficult data center temperature control and high energy consumption costs. Summary of the invention

[0004] The present invention provides a data center energy-saving cooling control method and system, which are used to solve the problem of high cooling energy consumption in existing data centers.

[0005] The present invention provides a data center energy-saving cooling control method, comprising: Obtaining the initial temperature setpoint and thermal safety temperature target of the air cooling device; Predicting the air cooling device power based on the initial temperature setting value using a pre-trained data center time series model; Under the constraint of the thermal safety temperature target, determining the ideal temperature setting value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer; The ideal temperature setting value is transmitted to a smoothing buffer, and a control instruction is sent to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device.

[0006] According to a data center energy-saving cooling control method provided by the present invention, the air cooling device power is predicted based on the initial temperature setting value by using a pre-trained data center time series model, specifically comprising: Predicting the average power of servers near the air cooling device within a set period using the data center time series model; Predicting the air inlet temperature of the air cooling device based on the average power of the server; The air inlet temperature of the air cooling device is compared with the initial temperature setting value to determine the temperature difference, and the power of the air cooling device is calculated according to the temperature difference.

[0007] According to a data center energy-saving cooling control method provided by the present invention, the average power of servers near an air cooling device is predicted by a data center time series model within a set period, specifically comprising: The total power of the server from the start time to the end time within the set period is predicted through the power prediction module in the data center time series model, and the average power of the server near the air cooling device is calculated based on the total power and the time parameters of the set period; Among them, the power prediction module in the data center time series model is obtained by training a preset first linear model based on the server power changes within a historical set period.

[0008] According to a data center energy-saving cooling control method provided by the present invention, the air inlet temperature of the air cooling device is compared with the initial temperature setting value to determine the temperature difference, and the power of the air cooling device is predicted according to the temperature difference, specifically comprising: Subtract the obtained air inlet temperature of the air cooling device from the initial temperature setting value to obtain a temperature difference; Based on the temperature difference, the power of the air cooling device is predicted by a cooling energy prediction module in a data center time series model; Among them, the cooling energy prediction module in the data center time series model is obtained by training a preset second linear model based on the cooling device power data corresponding to the historical temperature difference.

[0009] According to a data center energy-saving cooling control method provided by the present invention, the ideal temperature setting value corresponding to the minimum power value of the air cooling device is determined by a preset Bayesian optimizer under the constraint of the thermal safety temperature target, specifically comprising: Under the constraint of the thermal safety temperature target, error optimization is performed using a preset Bayesian optimizer to select the minimum value of the air cooling device power from the data set; wherein the data set is constructed by predicting the air cooling device power data under multiple temperature setting values ​​within a set range using a data center time series model; The corresponding ideal temperature setting value is determined according to the minimum value of the air cooling device power.

[0010] According to a data center energy-saving cooling control method provided by the present invention, the ideal temperature setting value is transmitted to a smoothing buffer, and a control instruction is sent to a controller of an air cooling device through the smoothing buffer to adjust the power of the air cooling device, specifically comprising: The ideal temperature setting value is transmitted to a smoothing buffer of a set length, and low-pass filtering is performed through the smoothing buffer to eliminate high-frequency changes and generate a control instruction for slow changes; The air cooling device power is adjusted based on the control command.

[0011] The present invention also provides a data center energy-saving cooling control system, the system comprising: A data acquisition module, used to obtain an initial temperature setting value and a thermal safety temperature target of the air cooling device; A prediction module, configured to predict the air cooling device power based on the initial temperature setting value through a pre-trained data center time series model; A set value output module, used to determine an ideal temperature set value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; The regulating module is used to transmit the ideal temperature setting value to the smoothing buffer, send a control instruction to the controller of the air cooling device through the smoothing buffer, and regulate the power of the air cooling device.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the data center energy-saving cooling control method as described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the data center energy-saving cooling control method as described above is implemented.

[0014] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned data center energy-saving cooling control methods.

[0015] The present invention provides a data center energy-saving cooling control method and system, which predicts the power of an air cooling device through a data center time series model, and can calculate the ideal temperature setting value corresponding to the minimum power of the air cooling device after Bayesian optimization according to the thermal safety temperature constraint, thereby reducing the cooling power consumption of the data center, and can slowly adjust the power of the air cooling device by setting a smoothing buffer to avoid sudden changes in the temperature value of the data center exceeding the thermal safety constraint. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a flow chart of a data center energy-saving cooling control method provided by the present invention.

[0018] Figure 2 It is a module connection diagram of a data center energy-saving cooling control system provided by the present invention.

[0019] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention.

[0020] Reference numerals: 110: data acquisition module; 120: prediction module; 130: set value output module; 140: adjustment module; 310: processor; 320: communication interface; 330: memory; 340: communication bus. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Combine the following Figure 1 A data center energy-saving cooling control method of the present invention is described, comprising: step 100, obtaining an initial temperature setting value and a thermal safety temperature target of an air cooling device.

[0023] In the present invention, the temperature variation of the data center and the associated air cooling device power at a given initial temperature set point are predicted within a limited time series range, and based on the prediction results, a temperature set point is selected to be executed at the current time step to minimize the predicted air cooling device power under the constraint that the predicted cold channel temperature captured by any sensor does not exceed the thermal safety temperature target. In addition to minimizing the air cooling device power, the optimization goal also includes minimizing cooling interruptions in turn to curb temperature rises that may violate thermal safety.

[0024] Step 200: predicting the air cooling device power based on the initial temperature setting value by using a pre-trained data center time series model.

[0025] In the present invention, the data center time series model uses four modules to predict the data center temperature and air cooling device power. In the first stage of the data center time series model, a power prediction module is established to predict the average server power of the next stage; in the second stage, the intake temperature prediction module is used to predict the actual intake temperature of the air cooling device according to the given initial temperature setting value and the output of the power prediction module; in the third stage, the power prediction module and the intake temperature module show the heat generated and discharged by the data center, and the temperature of the future data center captured by sensors on different racks is predicted by the data center temperature prediction module. In the fourth stage, for the prediction of the air cooling device power, a cooling energy prediction module is used to predict the air cooling device power with the predicted intake temperature and initial temperature setting value of the air cooling device as input.

[0026] Specifically, predicting the average power of servers near the air cooling device within a set period through the data center time series model; Predicting the air inlet temperature of the air cooling device based on the average power of the server; The air inlet temperature of the air cooling device is compared with the initial temperature setting value to determine the temperature difference, and the power of the air cooling device is calculated according to the temperature difference.

[0027] The average power of servers near the air cooling device is predicted by the data center time series model within a set period, including: The total power of the server from the start time to the end time within the set period is predicted through the power prediction module in the data center time series model, and the average power of the server near the air cooling device is calculated based on the total power and the time parameters of the set period; Among them, the power prediction module in the data center time series model is obtained by training a preset first linear model based on the server power changes within a historical set period.

[0028] The air inlet temperature of the air cooling device is compared with an initial temperature setting value to determine a temperature difference, and the power of the air cooling device is predicted according to the temperature difference, specifically including: Subtract the obtained air inlet temperature of the air cooling device from the initial temperature setting value to obtain a temperature difference; Based on the temperature difference, the power of the air cooling device is predicted by a cooling energy prediction module in a data center time series model; Among them, the cooling energy prediction module in the data center time series model is obtained by training a preset second linear model based on the cooling device power data corresponding to the historical temperature difference.

[0029] In the present invention, the average server power of the next stage is predicted by the power prediction module, and the server power represents the rate at which heat is generated inside the data center. However, the power consumption of each server may change suddenly, which is difficult to predict due to the unknown upper workload. To solve this problem, the average power aggregation of servers near the air cooling device is predicted according to the data center layout, because the possibility of multiple servers changing their power consumption at the same time is less than that of a single server.

[0030] Step 300: Under the constraint of the thermal safety temperature target, determine the ideal temperature setting value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer.

[0031] Specifically, under the constraint of the thermal safety temperature target, error optimization is performed by a preset Bayesian optimizer, and the minimum value of the air cooling device power is selected from the data set; wherein the data set is constructed by predicting the air cooling device power data under multiple temperature setting values ​​within a set range through a data center time series model; The corresponding ideal temperature setting value is determined according to the minimum value of the air cooling device power.

[0032] In the present invention, a data center time series model establishes how a given temperature set point affects the data center temperature at different locations in the data center within a limited time frame. Based on the model's predictions, an optimizer is used to find the energy optimal set point without violating the thermal safety temperature constraint. However, the optimizer needs to be aware of the modeling error of the data center time series model, although its generalization error has been reduced by incorporating the influence of exogenous inputs. The generalization error is only an average measure. Therefore, large deviations may still occur. When they occur, subsequent temperature set point optimizations are based on erroneous predictions, which makes the data center potentially vulnerable to thermal safety violations and suboptimal energy savings.

[0033] The following optimization problem is formed within the set period.

[0034] .

[0035] in, It means that the true objective function is To Output on step; Indicates that the true constraint function is To Output on step; is the objective function calculated according to the time series model in To Output on step; is the constraint function calculated according to the time series model in To Output on step; is the difference between the true target function output and the predicted value; is the difference between the true constraint function output and the predicted value; Set temperature values ​​for minimum / maximum air cooling; Indicated in The set temperature of the time step, other ones with different superscripts are replaced according to their positions; Indicates that the air cooling device is To Step 1 to set the temperature.

[0036] The Bayesian optimizer transforms the objective function Maximize, the objective function contains the constrained function The true cooling energy of the constraint, if the constraint function is positive, indicates a true thermal safety violation. However, only the predicted version and It can be directly obtained through the data center time series model. In order to avoid the energy impact of temperature set point changes, the Bayesian optimizer shares the same set point for all time steps in the time series, that is, . and are the minimum and maximum temperature setpoints achievable by the PID controller in an air cooling device.

[0037] Cooling outages at air cooling unit power and thermal safety temperature targets predicted by the data center time series model are incorporated into the optimization objectives.

[0038] Because cooling interruption will cause the temperature of the data center to rise rapidly. In addition, according to the existing data center cooling standards, the duration of data center cooling interruption is limited according to its availability specifications. Although cooling interruption provides an energy-saving incentive, it conflicts with this requirement. In the present invention, a Bayesian optimizer is selected to optimize the low-dimensional black box function (such as respectively from arrive and Based on the output of the data center time series model, the Bayesian optimizer samples values ​​from the distribution of the objective function and the constraint function. A modeling error-aware Bayesian optimizer is used to find the optimal feasible set value. In order to construct the distribution of the objective function and the constraint function, uncertainty estimates of the objective function and the constraint function are required. Using an online prediction error monitor, the prediction error of the data center time series model over the past day is tracked, which is a typical time period for data center load to rise and fall. Uncertainty estimates are obtained from the monitor using bootstrapping. After the Bayesian optimizer calculates a set point, it is sent to the smoothing buffer, whose output is ultimately executed by the PID controller of the air cooling unit.

[0039] Step 400: The ideal temperature setting value is transmitted to a smoothing buffer, and a control instruction is sent to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device.

[0040] Specifically, the method includes: transmitting the ideal temperature setting value to a smoothing buffer of a set length, performing low-pass filtering through the smoothing buffer to eliminate high-frequency changes and generate a control instruction for slow changes; and adjusting the power of the air cooling device based on the control instruction.

[0041] In the present invention, the Bayesian optimizer calculates the ideal temperature setting value, which needs to be processed and sent to the PID controller of the air cooling device for execution. The reason is that the conversion of the ideal temperature setting value takes time, which depends on the power of the server and the previous temperature setting value. Therefore, when the air cooling device has not stabilized near the previous few temperature setting values, the regular execution of the temperature setting value will inevitably result in some temperature setting values ​​being executed. This type of execution has the possibility of additional energy consumption. In order to avoid this situation, a smoothing buffer of a set length is added to store the ideal temperature setting value calculated by the Bayesian optimizer. Finally, it is sent to the PID controller of the air cooling device, and the ideal temperature setting value is the running average of the value stored in the smoothing buffer. The smoothing buffer acts as a low-pass filter, eliminating high-frequency changes in the calculated set point, thereby reducing unnecessary power peaks.

[0042] By providing an energy-saving cooling control method for a data center provided by the present invention, the power of an air cooling device is predicted by a time series model of a data center, and an ideal temperature setting value corresponding to the minimum power of the air cooling device can be calculated after Bayesian optimization according to a thermal safety temperature constraint, thereby reducing the cooling power consumption of the data center, and the power of the air cooling device can be slowly adjusted by setting a smoothing buffer, thereby avoiding sudden changes in the temperature value of the data center exceeding the thermal safety constraint.

[0043] The present invention also discloses a data center energy-saving cooling control system, the system comprising: A data acquisition module 110 is used to acquire an initial temperature setting value and a thermal safety temperature target of the air cooling device; A prediction module 120, configured to predict the air cooling device power based on the initial temperature setting value by using a pre-trained data center time series model; A set value output module 130, used to determine an ideal temperature set value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; The regulating module 140 is used to transmit the ideal temperature setting value to the smoothing buffer, and send a control instruction to the controller of the air cooling device through the smoothing buffer to regulate the power of the air cooling device.

[0044] The method of predicting the power of the air cooling device based on the initial temperature setting value by using a pre-trained data center time series model specifically includes: Predicting the average power of servers near the air cooling device within a set period using the data center time series model; Predicting the air inlet temperature of the air cooling device based on the average power of the server; The air inlet temperature of the air cooling device is compared with the initial temperature setting value to determine the temperature difference, and the power of the air cooling device is calculated according to the temperature difference.

[0045] The average power of servers near the air cooling device is predicted by the data center time series model within a set period, including: The total power of the server from the start time to the end time within the set period is predicted through the power prediction module in the data center time series model, and the average power of the server near the air cooling device is calculated based on the total power and the time parameters of the set period; Among them, the power prediction module in the data center time series model is obtained by training a preset first linear model based on the server power changes within a historical set period.

[0046] The air inlet temperature of the air cooling device is compared with an initial temperature setting value to determine a temperature difference, and the power of the air cooling device is predicted according to the temperature difference, specifically including: Subtract the obtained air inlet temperature of the air cooling device from the initial temperature setting value to obtain a temperature difference; Based on the temperature difference, the power of the air cooling device is predicted by a cooling energy prediction module in a data center time series model; Among them, the cooling energy prediction module in the data center time series model is obtained by training a preset second linear model based on the cooling device power data corresponding to the historical temperature difference.

[0047] Under the constraint of the thermal safety temperature target, the ideal temperature setting value corresponding to the minimum power value of the air cooling device is determined by a preset Bayesian optimizer, specifically including: Under the constraint of the thermal safety temperature target, error optimization is performed using a preset Bayesian optimizer to select the minimum value of the air cooling device power from the data set; wherein the data set is constructed by predicting the air cooling device power data under multiple temperature setting values ​​within a set range using a data center time series model; The corresponding ideal temperature setting value is determined according to the minimum value of the air cooling device power.

[0048] The ideal temperature setting value is transmitted to a smoothing buffer, and a control instruction is sent to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device, specifically including: The ideal temperature setting value is transmitted to a smoothing buffer of a set length, and low-pass filtering is performed through the smoothing buffer to eliminate high-frequency changes and generate a control instruction for slow changes; The air cooling device power is adjusted based on the control command.

[0049] A data center energy-saving cooling control method provided by the present invention predicts the power of an air cooling device through a data center time series model, and can calculate the ideal temperature setting value corresponding to the minimum power of the air cooling device after Bayesian optimization according to the thermal safety temperature constraint, thereby reducing the cooling power consumption of the data center, and can slowly adjust the power of the air cooling device by setting a smoothing buffer, thereby avoiding sudden changes in the temperature value of the data center that exceed the thermal safety constraint.

[0050] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the data center energy-saving cooling control method, which includes: obtaining the initial temperature setting value and the thermal safety temperature target of the air cooling device; predicting the power of the air cooling device through the pre-trained data center time series model based on the initial temperature setting value; determining the ideal temperature setting value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; transmitting the ideal temperature setting value to the smoothing buffer, and sending a control instruction to the controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device.

[0051] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0052] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the data center energy-saving cooling control method provided by the above methods, and the method includes: obtaining an initial temperature setting value and a thermal safety temperature target of an air cooling device; predicting the power of the air cooling device through a pre-trained data center time series model based on the initial temperature setting value; determining an ideal temperature setting value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; transmitting the ideal temperature setting value to a smoothing buffer, and sending a control instruction to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device.

[0053] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the data center energy-saving cooling control method provided by the above-mentioned methods, the method comprising: obtaining an initial temperature setting value and a thermal safety temperature target of an air cooling device; predicting the power of the air cooling device through a pre-trained data center time series model based on the initial temperature setting value; determining an ideal temperature setting value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; transmitting the ideal temperature setting value to a smoothing buffer, and sending a control instruction to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device.

[0054] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0055] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0056] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data center energy-saving cooling control method, characterized in that: include: Obtaining the initial temperature setpoint and thermal safety temperature target of the air cooling device; Predicting the air cooling device power based on the initial temperature setting value using a pre-trained data center time series model; Under the constraint of the thermal safety temperature target, determining the ideal temperature setting value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer; The ideal temperature setting value is transmitted to a smoothing buffer, and a control instruction is sent to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device.

2. The data center energy-saving cooling control method according to claim 1, characterized in that: The predicting of the air cooling device power based on the initial temperature setting value by using a pre-trained data center time series model specifically includes: Predicting the average power of servers near the air cooling device within a set period using the data center time series model; Predicting the air inlet temperature of the air cooling device based on the average power of the server; The air inlet temperature of the air cooling device is compared with the initial temperature setting value to determine the temperature difference, and the power of the air cooling device is calculated according to the temperature difference.

3. The data center energy-saving cooling control method according to claim 2, characterized in that: The predicting the average power of the servers near the air cooling device by using the data center time series model within the set period specifically includes: The total power of the server from the start time to the end time within the set period is predicted through the power prediction module in the data center time series model, and the average power of the server near the air cooling device is calculated based on the total power and the time parameters of the set period; Among them, the power prediction module in the data center time series model is obtained by training a preset first linear model based on the server power changes within a historical set period.

4. The data center energy-saving cooling control method according to claim 2, characterized in that: The step of comparing the air inlet temperature of the air cooling device with the initial temperature setting value to determine the temperature difference, and predicting the power of the air cooling device according to the temperature difference specifically includes: Subtract the obtained air inlet temperature of the air cooling device from the initial temperature setting value to obtain a temperature difference; Based on the temperature difference, the power of the air cooling device is predicted by a cooling energy prediction module in a data center time series model; Among them, the cooling energy prediction module in the data center time series model is obtained by training a preset second linear model based on the cooling device power data corresponding to the historical temperature difference.

5. The data center energy-saving cooling control method according to claim 1, characterized in that: The method of determining the ideal temperature setting value corresponding to the minimum power value of the air cooling device by a preset Bayesian optimizer under the constraint of the thermal safety temperature target specifically includes: Under the constraint of the thermal safety temperature target, error optimization is performed using a preset Bayesian optimizer to select the minimum value of the air cooling device power from the data set; wherein the data set is constructed by predicting the air cooling device power data under multiple temperature setting values ​​within a set range using a data center time series model; The corresponding ideal temperature setting value is determined according to the minimum value of the air cooling device power.

6. The data center energy-saving cooling control method according to claim 1, characterized in that: The ideal temperature setting value is transmitted to a smoothing buffer, and a control instruction is sent to a controller of the air cooling device through the smoothing buffer to adjust the power of the air cooling device, specifically including: The ideal temperature setting value is transmitted to a smoothing buffer of a set length, and low-pass filtering is performed through the smoothing buffer to eliminate high-frequency changes and generate a control instruction for slow changes; The air cooling device power is adjusted based on the control command.

7. A data center energy-saving cooling control system, characterized in that: The system comprises: A data acquisition module, used to obtain an initial temperature setting value and a thermal safety temperature target of the air cooling device; A prediction module, configured to predict the air cooling device power based on the initial temperature setting value through a pre-trained data center time series model; A set value output module, used to determine an ideal temperature set value corresponding to the minimum power value of the air cooling device through a preset Bayesian optimizer under the constraint of the thermal safety temperature target; The regulating module is used to transmit the ideal temperature setting value to the smoothing buffer, and send a control instruction to the controller of the air cooling device through the smoothing buffer to regulate the power of the air cooling device.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the data center energy-saving cooling control method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the data center energy-saving cooling control method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the data center energy-saving cooling control method according to any one of claims 1 to 6 is implemented.

Images