Control method, device and computer readable storage medium of air conditioner

Abstract

This disclosure relates to a control method, apparatus, and computer-readable storage medium for an air conditioner, and pertains to the field of computer technology. The method includes: acquiring current environmental state information of a data center server room; inputting the current environmental state information into a feature extraction module of an air conditioner control model to obtain current environmental features, wherein the air conditioner control model includes a feature extraction module and multiple decision modules, each decision module corresponding to different operating modes and control methods, including ice storage mode and ice melting mode, and return air control mode and supply air control mode; selecting a decision module corresponding to the current operating mode and current control method of the air conditioner; and inputting the current environmental features into the selected decision module to obtain control parameters for the air conditioner.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a control method, apparatus and computer-readable storage medium for an air conditioner. Background Technology

[0002] Currently, there are a huge number of data centers. In order to ensure the normal operation of servers in data center computer rooms, the energy consumption of cooling equipment is in the hundreds of millions every year.

[0003] Controlling the air conditioning in a data center to ensure the normal operation of the server room, such as optimizing airflow organization and adjusting the return air temperature, requires professionals to make multiple adjustments, which is labor-intensive, inflexible, and lacks precision. Summary of the Invention

[0004] In view of this, one of the technical problems to be solved by this disclosure is: how to improve the efficiency and accuracy of air conditioning control in data centers.

[0005] According to some embodiments of this disclosure, an air conditioning control method is provided, comprising: acquiring current environmental status information of a data center computer room; inputting the current environmental status information into a feature extraction module of an air conditioning control model to obtain current environmental features, wherein the air conditioning control model includes a feature extraction module and multiple decision modules, each decision module corresponding to different operating modes and control methods, the operating modes including ice storage mode and ice melting mode, and the control methods including return air control method and supply air control method; selecting a decision module corresponding to the current operating mode and current control method of the air conditioner; and inputting the current environmental features into the selected decision module to obtain control parameters of the air conditioner.

[0006] In some embodiments, the method further includes: generating multiple sets of training data for each combination of operating mode and control method, each set of training data including: environmental state information and corresponding labeled control parameters of the air conditioner; training the feature extraction module using all the training data; and training the decision module corresponding to the combination of operating mode and control method using the training data corresponding to each combination of operating mode and control method.

[0007] In some embodiments, generating multiple sets of training data for each combination of operating mode and control method includes: collecting real environmental state information and corresponding air conditioning control parameters in the data center computer room for each combination of operating mode and control method as partial training data; setting environmental state information in the data center computer room as simulated environmental state information; using the simulated environmental state information and energy consumption simulator to obtain the air conditioning control parameters as simulated control parameters of the air conditioning, and using the simulated environmental state information and corresponding simulated control parameters as partial training data.

[0008] In some embodiments, obtaining the control parameters of the air conditioner using an energy consumption simulator includes: taking the simulated environmental state information as the environmental state information at the current moment, and inputting it along with preset control parameters of the air conditioner for multiple consecutive moments starting from the current moment into the energy consumption simulator to obtain the environmental state information and power utilization efficiency (PUE) of the data center at each moment after the current moment output by the energy consumption simulator; determining the loss function value based on the data center PUE at each moment starting from the current moment, wherein the loss function value is positively correlated with the data center PUE; updating the preset control parameters of the air conditioner for multiple consecutive moments starting from the current moment; repeating the above steps until the loss function value is minimized to obtain the corresponding control parameters of the air conditioner for multiple consecutive moments starting from the current moment.

[0009] In some embodiments, the energy consumption simulator outputs environmental state information at each time point after the current time. The environmental state information includes at least one of the temperature and humidity of the data center rack. Determining the loss function value based on the data center PUE at each time point starting from the current time includes: determining the loss function value based on the data center PUE at each time point starting from the current time, and the difference between the environmental state information at each time point starting from the current time and the preset environmental state information. The loss function value is positively correlated with the data center PUE and negatively correlated with the difference between the environmental state information at each time point and the preset environmental state information.

[0010] In some embodiments, training the feature extraction module using all training data, and training the decision module corresponding to each combination of operating mode and control method using training data, includes: inputting the environmental state information of the training data into the feature extraction module of the air conditioning control model to obtain environmental features; selecting a decision module corresponding to the operating mode and control method of the training data; inputting the environmental features into the selected decision module to obtain the output air conditioning control parameters; determining the loss function of the air conditioning control model based on the output air conditioning control parameters and the labeled air conditioning control parameters; adjusting the parameters of the feature extraction module and the selected decision module based on the loss function of the air conditioning control model; repeating the above steps until a preset convergence condition is reached to complete the training.

[0011] In some embodiments, the environmental status information includes at least one of the following: temperature of the server room rack, humidity of the server room rack, IT load, cold aisle temperature, cold aisle humidity, hot aisle temperature, hot aisle humidity, real-time fan speed of the air conditioner, real-time return air temperature of the air conditioner, outdoor temperature, and outdoor humidity; the control parameters of the air conditioner include at least one of the following: set temperature of the air conditioner, maximum fan power of the air conditioner, minimum fan power of the air conditioner, and water valve opening.

[0012] According to some other embodiments of this disclosure, an air conditioning control device is provided, comprising: an acquisition module for acquiring current environmental status information of a data center computer room; a control module for inputting the current environmental status information into a feature extraction module of an air conditioning control model to obtain current environmental features, wherein the air conditioning control model includes a feature extraction module and multiple decision modules, each decision module corresponding to different operating modes and control methods, the operating modes including ice storage mode and ice melting mode, and the control methods including return air control method and supply air control method; selecting a decision module corresponding to the current operating mode and current control method of the air conditioner; and inputting the current environmental features into the selected decision module to obtain control parameters of the air conditioner.

[0013] In some embodiments, the apparatus further includes: a generation module, configured to generate multiple sets of training data for each combination of operating mode and control method, each set of training data including: environmental state information and corresponding labeled control parameters of the air conditioner; and a training module, configured to train the feature extraction module using all the training data, and train the decision module corresponding to the combination of operating mode and control method using the training data corresponding to each combination of operating mode and control method.

[0014] In some embodiments, the generation module is used to collect real environmental state information and corresponding air conditioning control parameters in the data center computer room for each combination of operating mode and control method, as part of the training data; set environmental state information in the data center computer room as simulated environmental state information; use the simulated environmental state information and energy consumption simulator to obtain the air conditioning control parameters as simulated control parameters of the air conditioning, and use the simulated environmental state information and corresponding simulated control parameters as part of the training data.

[0015] In some embodiments, the generation module is used to input the simulated environmental state information as the environmental state information at the current moment, along with preset control parameters for the air conditioners at multiple consecutive moments starting from the current moment, into the energy consumption simulator to obtain the environmental state information and power utilization efficiency (PUE) of the data center at each moment after the current moment output by the energy consumption simulator; determine the loss function value based on the data center PUE at each moment starting from the current moment, wherein the loss function value is positively correlated with the data center PUE; update the preset control parameters for the air conditioners at multiple consecutive moments starting from the current moment; repeat the above steps until the loss function value is minimized to obtain the corresponding control parameters for the air conditioners at multiple consecutive moments starting from the current moment.

[0016] According to some other embodiments of this disclosure, an air conditioner control device is provided, comprising: a processor; and a memory coupled to the processor for storing instructions, which, when executed by the processor, cause the processor to perform an air conditioner control method as described in any of the foregoing embodiments.

[0017] According to further embodiments of the present disclosure, a non-transitory computer-readable storage medium is provided, on which a computer program is stored, wherein the program, when executed by a processor, implements the air conditioning control method of any of the foregoing embodiments.

[0018] This disclosure inputs the current environmental status information of the data center server room into an air conditioning control model. The air conditioning control model includes a feature extraction module and multiple decision modules, which can select the appropriate decision modules for different operating conditions and control methods to output the air conditioning control parameters. Through the air conditioning control model, more accurate control parameters can be automatically obtained for the current environment, improving the control efficiency and accuracy of the air conditioning. Furthermore, because different decision modules are designed for different operating conditions and control methods in the air conditioning control model, it can be applied to various situations, further improving the accuracy of air conditioning control.

[0019] Other features and advantages of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart illustrating an air conditioner control method according to some embodiments of the present disclosure is shown.

[0022] Figure 2 A flowchart illustrating an air conditioning control method according to other embodiments of this disclosure is shown.

[0023] Figure 3 A schematic diagram illustrating an air conditioner control method according to some embodiments of the present disclosure is shown.

[0024] Figure 4 A schematic diagram of the structure of an air conditioner control device according to some embodiments of the present disclosure is shown.

[0025] Figure 5 A schematic diagram of the structure of an air conditioner control device according to other embodiments of the present disclosure is shown.

[0026] Figure 6 A schematic diagram of the structure of an air conditioner control device according to further embodiments of the present disclosure is shown. Detailed Implementation

[0027] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit this disclosure or its application or use. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0028] This disclosure provides a method for controlling an air conditioner, which will be described below in conjunction with... Figures 1-3 Describe it.

[0029] Figure 1 Flowcharts are shown for some embodiments of the air conditioning control method disclosed herein. For example... Figure 1 As shown, the method of this embodiment includes steps S102 to S108.

[0030] In step S102, the current environmental status information of the data center computer room is obtained.

[0031] In some embodiments, environmental status information includes at least one of the following: temperature of the server room rack, humidity of the server room rack, IT load, cold aisle temperature, cold aisle humidity, hot aisle temperature, hot aisle humidity, real-time fan speed of the air conditioner, real-time return air temperature of the air conditioner, outdoor temperature, and outdoor humidity. Other environmental status information can also be set as needed, and is not limited to the examples given.

[0032] In step S104, the current environmental state information is input into the feature extraction module of the air conditioning control model to obtain the current environmental features.

[0033] The air conditioning control model can be pre-trained, and the training process will be described later. The air conditioning control model may include a feature extraction module and multiple decision modules, each corresponding to different operating modes and control methods. Operating modes may include, for example, ice storage mode and ice melting mode; control methods may include, for example, return air control and supply air control. The operating modes and control methods can be set according to the actual operation of the air conditioner and are not limited to the examples given. For instance, if there are two operating modes and two control methods, there would be four decision modules, corresponding to ice storage + return air control, ice storage + supply air control, ice melting + return air control, and ice melting + supply air control, respectively.

[0034] The feature extraction module may be an MLP (Multilayer Perceptron), but is not limited to the examples given. The decision-making module may be a classifier or regressor, selected based on the actual output data format. The overall air conditioning control model may be a neural network model.

[0035] In step S106, a decision module corresponding to the current operating mode and current control method of the air conditioner is selected.

[0036] The air conditioning control model can automatically select the decision module corresponding to the current operating mode and control method. For example, data representing the current operating mode and control method can be input along with current environmental status information, and the air conditioning control model will automatically select the corresponding decision module based on this data.

[0037] In step S108, the current environmental characteristics are input into the selected decision module to obtain the control parameters of the air conditioner.

[0038] In some embodiments, the control parameters of the air conditioner include at least one of the following: the set temperature of the air conditioner, the maximum power of the air conditioner's fan, the minimum power of the air conditioner's fan, and the water valve opening degree. The control parameters can be determined based on the actual adjustable parameters of the air conditioner, and are not limited to the examples given. Adjusting the air conditioner according to the output control parameters can better adapt to the current environmental conditions and ensure the normal operation of the computer room.

[0039] The method described in the above embodiment inputs the current environmental status information of the data center server room into the air conditioning control model. The air conditioning control model includes a feature extraction module and multiple decision modules, which can select the corresponding decision modules for different operating conditions and control methods to output the air conditioning control parameters. Through the air conditioning control model, more accurate control parameters can be automatically obtained for the current environment, improving the control efficiency and accuracy of the air conditioning. Furthermore, since different decision modules are designed for different operating conditions and control methods in the air conditioning control model, it can be applied to various situations, further improving the accuracy of air conditioning control.

[0040] The following is combined with Figure 2 Describe the training method for the air conditioning control model.

[0041] Figure 2 Flowcharts are shown for some other embodiments of the air conditioning control method disclosed herein. For example... Figure 2 As shown, the method of this embodiment includes steps S202 to S204.

[0042] In step S202, multiple sets of training data are generated for each combination of operating mode and control method.

[0043] Each set of training data includes, for example, environmental condition information and corresponding labeled air conditioning control parameters. For instance, training data is generated for four combinations: ice storage + return air control, ice storage + supply air control, ice melting + return air control, and ice melting + supply air control. For each combination of operating mode and control method, real environmental condition information and corresponding air conditioning control parameters from the data center can be collected as training data. That is, environmental condition information during actual operation of the data center and the actual air conditioning control parameters corresponding to each environmental condition are collected as training data.

[0044] The inventors discovered that because the environment in the computer room is constantly controlled by maintenance personnel, data with large fluctuations or extreme conditions is lacking, such as extremely cold (<10 degrees Celsius) or extremely hot (>30 degrees Celsius) computer room environments and control parameters. This data plays a crucial role in model training, improving the model's generalization ability and accuracy. Acquiring and labeling this data is difficult; therefore, this disclosure also proposes a method for automatically generating training data during the training process.

[0045] In some embodiments, for each combination of operating mode and control method, real environmental state information in the data center computer room and corresponding air conditioning control parameters are collected as partial training data; environmental state information in the data center computer room is set as simulated environmental state information; using the simulated environmental state information and energy consumption simulator, the control parameters of the air conditioner are obtained as simulated control parameters of the air conditioner, and the simulated environmental state information and corresponding simulated control parameters are used as partial training data.

[0046] Environmental state information can be manually set as simulated environmental state information, which can include relatively extreme data. Energy consumption simulators can be existing energy consumption simulation software, such as Energy Plus. The energy consumption simulator can take the current environmental state information and the current air conditioning control parameters as input, and output the environmental state information and the data center PUE (Power Usage Effectiveness) for the next moment.

[0047] Furthermore, in some embodiments, the simulated environmental state information is used as the environmental state information at the current moment, and is input into the energy consumption simulator along with the preset control parameters of the air conditioner for multiple consecutive moments starting from the current moment. This yields the environmental state information and data center PUE at each moment after the current moment, output by the energy consumption simulator. Based on the data center PUE at each moment starting from the current moment, the loss function value is determined. The preset control parameters of the air conditioner for multiple consecutive moments starting from the current moment are updated. The above steps are repeated until the loss function value is minimized, thus obtaining the corresponding control parameters of the air conditioner for multiple consecutive moments starting from the current moment.

[0048] For example, the environmental state information at time t is represented as x. t The control parameters of the air conditioner at time t are expressed as u. t u t The control parameters of the air conditioner to be set at time t, i.e., the action at time t, are executed by u. t Then it enters the environment state t+1. The dynamic model of the environment can be expressed by the following formula.

[0049] x t+1 =f(x) t u t (1)

[0050] In formula (1), x t and x t+1 These represent the environmental state information at times t and t+1, respectively, u t Let be the air conditioning control parameters at time t.

[0051] Furthermore, the first-order approximation of formula (1) is obtained using an energy consumption simulator, as shown in formula (2).

[0052]

[0053] In formula (2), F t [·] represents the function of the energy consumption simulator, f t It is a constant.

[0054] The loss function value is positively correlated with the data center's PUE. For example, the loss function can be expressed as the following formula.

[0055]

[0056] In formula (3), n is a preset value, and c(x) t u t ) represents the computer room PUE at time t, output by the energy consumption simulator.

[0057] Set the simulated environment state information at a given time, for example, as x1, the F of the energy consumption simulator. t [·] Given that we need to find u1, ..., u at each time point that satisfies formula (3). t You can first set u1, ..., u randomly or based on experience. t , let u1, ..., u t Input x1 into the energy consumption simulator and obtain c(x) at each time step. t u t ), calculate the loss function value, and update u1, ..., u t Again, u1, ..., u tInput x1 into the energy consumption simulator, calculate the loss function value, and repeat the above steps until the loss function value is minimized, thus obtaining the corresponding u1, ..., u t We can also obtain x1, ..., x t x1, ..., x t and u1, ..., u t As training data, u1, ..., u t Using control parameters that minimize energy consumption as training data allows the air conditioning control model to learn the optimal control parameters, enabling the model to output control parameters that minimize energy consumption during application.

[0058] In some embodiments, a loss function value is determined based on the data center PUE at each time starting from the current time, and the difference between the environmental state information at each time starting from the current time and the preset environmental state information. The loss function value is positively correlated with the data center PUE and negatively correlated with the difference between the environmental state information at each time and the preset environmental state information.

[0059] For example, the loss function value is positively correlated with the PUE of the data center, and negatively correlated with the difference between the temperature of the data center's front-end cabinets at various times and the preset temperature, and negatively correlated with the difference between the humidity of the data center's front-end cabinets at various times and the preset humidity. For example, c(x t u t ) is represented by the following formula.

[0060]

[0061] In formula (4), α1, α2, and α3 are weights. This represents the temperature of the front-end cabinet in the computer room at time t. H represents the humidity of the server rack in the computer room at time t, where T is the preset temperature and H is the preset humidity.

[0062] The method described in the above embodiments can simulate multiple sets of training data, including extreme data, and realize the automatic labeling and generation of data. The generated training data is data under energy-saving mode, that is, the energy consumption value is minimized or reduced. Using this training data to train the air conditioning control model, the model can learn the air conditioning control mode under energy-saving mode.

[0063] In step S204, the feature extraction module is trained using all the training data, and the decision module corresponding to the combination of each working condition mode and control method is trained using the training data corresponding to each combination of working condition mode and control method.

[0064] In some embodiments, the environmental state information of the training data is input into the feature extraction module of the air conditioning control model to obtain environmental features; a decision module corresponding to the operating mode and control method of the training data is selected; the environmental features are input into the selected decision module to obtain the output air conditioning control parameters; the loss function of the air conditioning control model is determined based on the output air conditioning control parameters and the labeled air conditioning control parameters; the parameters of the feature extraction module and the selected decision module are adjusted according to the loss function of the air conditioning control model; the above steps are repeated until the preset convergence condition is reached, and the training is completed. The specific training process can refer to existing neural network training processes and will not be elaborated further. Preset convergence conditions include, for example, reaching a preset number of iterations, minimizing the loss function value, or the loss function value being less than a certain threshold.

[0065] like Figure 3 As shown, the model disclosed herein comprises two parts: a model for generating training data and an air conditioning control model. The training data includes real data from a normal computer room and data generated using an energy consumption simulator. The real data from the normal computer room can be directly labeled, while the data generated by the energy consumption simulator includes extreme case data for data augmentation, labeled using the simulator's output. The training data is used to train the air conditioning control model. The air conditioning control model includes a feature extraction module and multiple decision modules. Environmental state information is input to the feature extraction module, which outputs environmental features. A selection module selects the decision module corresponding to the operating mode and control method, and the decision module outputs the air conditioning control parameters.

[0066] The method described in the above embodiments uses an energy consumption simulator to generate extreme case data and performs automated annotation, correcting the problems of missing and imbalanced data and significantly improving the efficiency of data annotation. This data is then used for subsequent model training to improve the model's accuracy and generalization. The air conditioning control model integrates multiple operating conditions and control methods, shares parameters from the feature extraction part, and uses conditional selection to distinguish the decision parts of different operating conditions and control methods, greatly improving the model's generalization ability and efficiency, and enhancing its robustness.

[0067] This disclosure also provides an air conditioning control device, which is described below in conjunction with... Figure 4 Describe it.

[0068] Figure 4 This is a structural diagram of some embodiments of the air conditioning control device disclosed herein. For example... Figure 4 As shown, the device 40 in this embodiment includes: an acquisition module 410 and a control module 420.

[0069] The acquisition module 410 is used to acquire the current environmental status information of the data center computer room.

[0070] The control module 420 is used to input the current environmental state information into the feature extraction module of the air conditioning control model to obtain the current environmental features. The air conditioning control model includes a feature extraction module and multiple decision modules. Each decision module corresponds to a different operating mode and control method. The operating modes include ice storage mode and ice melting mode, and the control methods include return air control mode and supply air control mode. The control module is selected to correspond to the current operating mode and current control method of the air conditioner. The current environmental features are input into the selected decision module to obtain the control parameters of the air conditioner.

[0071] In some embodiments, the device 40 further includes a generation module 430, which generates multiple sets of training data for each combination of operating mode and control method, each set of training data including environmental state information and corresponding labeled control parameters of the air conditioner.

[0072] In some embodiments, the generation module 430 is used to collect real environmental state information and corresponding air conditioning control parameters in the data center computer room for each combination of operating mode and control method, as part of the training data; set environmental state information in the data center computer room as simulated environmental state information; and use the simulated environmental state information and energy consumption simulator to obtain the air conditioning control parameters as simulated control parameters of the air conditioning, and use the simulated environmental state information and corresponding simulated control parameters as part of the training data.

[0073] In some embodiments, the generation module 430 is used to input the simulated environmental state information as the environmental state information at the current moment, along with preset control parameters for the air conditioner for multiple consecutive moments starting from the current moment, into the energy consumption simulator to obtain the environmental state information and power utilization efficiency (PUE) of the computer room at each moment after the current moment output by the energy consumption simulator; determine the loss function value based on the PUE of the computer room at each moment starting from the current moment, wherein the loss function value is positively correlated with the PUE of the computer room; update the preset control parameters for the air conditioner for multiple consecutive moments starting from the current moment; repeat the above steps until the loss function value is minimized to obtain the corresponding control parameters for the air conditioner for multiple consecutive moments starting from the current moment.

[0074] Training module 440 is used to train the feature extraction module using all the training data, and to train the decision module corresponding to the combination of each working condition mode and control method using the training data corresponding to each combination of working condition mode and control method.

[0075] In some embodiments, the training module 440 is used to input the environmental state information of the training data into the feature extraction module of the air conditioning control model to obtain environmental features; select a decision module corresponding to the operating mode and control method of the training data; input the environmental features into the selected decision module to obtain the output air conditioning control parameters; determine the loss function of the air conditioning control model based on the output air conditioning control parameters and the labeled air conditioning control parameters; adjust the parameters of the feature extraction module and the selected decision module based on the loss function of the air conditioning control model; repeat the above steps until the preset convergence condition is reached to complete the training.

[0076] In some embodiments, the environmental status information includes at least one of the following: temperature of the server room rack, humidity of the server room rack, IT load, cold aisle temperature, cold aisle humidity, hot aisle temperature, hot aisle humidity, real-time fan speed of the air conditioner, real-time return air temperature of the air conditioner, outdoor temperature, and outdoor humidity; the control parameters of the air conditioner include at least one of the following: set temperature of the air conditioner, maximum fan power of the air conditioner, minimum fan power of the air conditioner, and water valve opening.

[0077] The air conditioning control device in the embodiments of this disclosure can be implemented by various computing devices or computer systems, as described below. Figure 5 as well as Figure 5 Describe it.

[0078] Figure 5 This is a structural diagram of some embodiments of the air conditioning control device disclosed herein. For example... Figure 5 As shown, the apparatus 50 of this embodiment includes a memory 510 and a processor 520 coupled to the memory 510. The processor 520 is configured to execute the air conditioning control method of any of the embodiments of this disclosure based on instructions stored in the memory 510.

[0079] The memory 510 may include, for example, system memory, fixed non-volatile storage media, etc. The system memory may store, for example, an operating system, application programs, a boot loader, a database, and other programs.

[0080] Figure 6 Structural diagrams of other embodiments of the air conditioning control device disclosed herein are shown. Figure 6As shown, the device 60 in this embodiment includes a memory 610 and a processor 620, which are similar to the memory 510 and processor 520, respectively. It may also include an input / output interface 630, a network interface 640, a storage interface 650, etc. These interfaces 630, 640, 650, and the memory 610 and processor 620 can be connected, for example, via a bus 660. The input / output interface 630 provides a connection interface for input / output devices such as a display, mouse, keyboard, and touchscreen. The network interface 640 provides a connection interface for various networked devices, such as connecting to a database server or cloud storage server. The storage interface 650 provides a connection interface for external storage devices such as SD cards and USB flash drives.

[0081] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable non-transitory storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0082] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0083] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0084] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0085] The above description is only a preferred embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A method for controlling an air conditioner, comprising: For each combination of operating mode and control method, real environmental state information and corresponding air conditioning control parameters in the data center computer room are collected as part of the training data. Environmental state information in the data center computer room is set as simulated environmental state information. Using the simulated environmental state information and energy consumption simulator, the control parameters of the air conditioner are obtained as simulated control parameters of the air conditioner. The simulated environmental state information and the corresponding simulated control parameters are used as part of the training data to generate multiple sets of training data. Each set of training data includes environmental state information and the corresponding labeled control parameters of the air conditioner. The feature extraction module of the air conditioning control model is trained using all the training data, and the decision module corresponding to the combination of each operating mode and control method is trained using the training data corresponding to each combination of operating mode and control method. Obtain the current environmental status information of the data center server room; The current environmental state information is input into the feature extraction module of the air conditioning control model to obtain the current environmental features. The air conditioning control model includes a feature extraction module and multiple decision modules. Each decision module corresponds to a different operating mode and control method. The operating mode includes ice storage mode and ice melting mode. The control method includes return air control method and air supply control method. Select the decision module corresponding to the current operating mode and current control method of the air conditioner; The current environmental characteristics are input into the selected decision module to obtain the control parameters of the air conditioner.

2. The control method according to claim 1, wherein, The method of obtaining the control parameters of the air conditioner using an energy consumption simulator includes: The simulated environmental state information is used as the environmental state information at the current moment, and the control parameters of the air conditioner for multiple consecutive moments starting from the current moment are input into the energy consumption simulator to obtain the environmental state information and power usage efficiency (PUE) of the computer room at each moment after the current moment output by the energy consumption simulator. The loss function value is determined based on the data center PUE at each time starting from the current time, wherein the loss function value is positively correlated with the data center PUE; Update the preset control parameters of the air conditioner for multiple consecutive time periods starting from the current time; Repeat the above steps until the loss function value is minimized, and obtain the corresponding control parameters of the air conditioner for multiple consecutive times starting from the current time.

3. The control method according to claim 2, wherein, The energy consumption simulator outputs environmental status information for each time after the current time, and the environmental status information includes at least one of the temperature and humidity of the computer room rack. The step of determining the loss function value based on the data center PUE at each time point starting from the current time includes: The loss function value is determined based on the data center PUE at each time starting from the current time, and the difference between the environmental state information at each time starting from the current time and the preset environmental state information. The loss function value is positively correlated with the data center PUE and negatively correlated with the difference between the environmental state information at each time and the preset environmental state information.

4. The control method according to claim 1, wherein, The step of training the feature extraction module using all the training data, and training the decision module corresponding to each combination of operating conditions and control methods using the training data corresponding to each combination of operating conditions and control methods, includes: The environmental state information of the training data is input into the feature extraction module of the air conditioning control model to obtain environmental features; Select a decision module that corresponds to the operating condition mode and control method of the training data; The environmental characteristics are input into the selected decision module to obtain the output control parameters of the air conditioner; Based on the output control parameters of the air conditioner and the labeled control parameters of the air conditioner, determine the loss function of the air conditioner control model; The parameters of the feature extraction module and the parameters of the selected decision module are adjusted according to the loss function of the air conditioning control model. Repeat the above steps until the preset convergence condition is met to complete the training.

5. The control method according to any one of claims 1-4, wherein, The environmental status information includes at least one of the following: temperature of the server rack, humidity of the server rack, IT load, cold aisle temperature, cold aisle humidity, hot aisle temperature, hot aisle humidity, real-time fan speed of the air conditioner, real-time return air temperature of the air conditioner, outdoor temperature, and outdoor humidity. The control parameters of the air conditioner include at least one of the following: the set temperature of the air conditioner, the maximum power of the air conditioner's fan, the minimum power of the air conditioner's fan, and the water valve opening degree.

6. A control device for an air conditioner, comprising: The generation module is used to collect real environmental state information and corresponding air conditioning control parameters in the data center computer room for each combination of operating mode and control method, as part of the training data; set the environmental state information in the data center computer room as simulated environmental state information; use the simulated environmental state information and energy consumption simulator to obtain the control parameters of the air conditioner as simulated control parameters of the air conditioner; and use the simulated environmental state information and corresponding simulated control parameters as part of the training data to generate multiple sets of training data, wherein each set of training data includes environmental state information and corresponding labeled control parameters of the air conditioner. The training module is used to train the feature extraction module of the air conditioning control model using all the training data, and to train the decision module corresponding to the combination of each operating mode and control method using the training data corresponding to each combination of operating mode and control method. The acquisition module is used to acquire the current environmental status information of the data center computer room; The control module is used to input the current environmental state information into the feature extraction module of the air conditioning control model to obtain the current environmental features. The air conditioning control model includes one feature extraction module and multiple decision modules, each corresponding to a different operating mode and control method. The operating modes include ice storage mode and ice melting mode, and the control methods include return air control mode and supply air control mode. The module corresponding to the current operating mode and control method of the air conditioner is selected. The current environmental features are input into the selected decision module to obtain the control parameters of the air conditioner.

7. The control device according to claim 6, wherein, The generation module is used to take the simulated environmental state information as the environmental state information at the current moment, and input it into the energy consumption simulator along with the preset control parameters of the air conditioner for multiple consecutive moments starting from the current moment. This yields the environmental state information and power usage efficiency (PUE) of the data center at each moment after the current moment, output by the energy consumption simulator. Based on the data center PUE at each moment starting from the current moment, a loss function value is determined, wherein the loss function value is positively correlated with the data center PUE. The preset control parameters of the air conditioner for multiple consecutive moments starting from the current moment are updated. The above steps are repeated until the loss function value is minimized, thus obtaining the corresponding control parameters of the air conditioner for multiple consecutive moments starting from the current moment.

8. A control device for an air conditioner, comprising: processor; as well as A memory coupled to the processor is used to store instructions that, when executed by the processor, cause the processor to perform the air conditioning control method as described in any one of claims 1-5.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, wherein, When executed by a processor, the program implements the steps of the method according to any one of claims 1-5.

Images