Control system, control method, and program

Abstract

This control system is provided with: an acquisition unit for acquiring, in a prescribed cycle, prescribed information related to a plurality of information technology devices installed in a space, one or more air-cooling devices for air-cooling the space and at least one of the information technology devices, one or more liquid-cooling devices for liquid-cooling at least one of the information technology devices, and the space; a prediction unit for predicting a prescribed future state in the space on the basis of the information by using a prediction model obtained by virtually reproducing the space on a computer; and a control unit for controlling at least one of the information technology devices, the air-cooling devices, and the liquid-cooling devices on the basis of the future state.

Description

Control System, Control Method, and Program

[0001] This disclosure relates to a control system, a control method, and a program. This application claims priority based on Japanese Patent Application No. 2024-232662 filed in Japan on December 27, 2024, and incorporates its content herein by reference.

[0002] Patent Document 1 discloses a method for determining the positions of devices arranged in a server room so as to satisfy the physical conditions and the thermal fluid dynamic conditions in the server room.

[0003] Japanese Unexamined Patent Application Publication No. 2015-114768

[0004] Generally, devices arranged in a server room require appropriate control during operation. However, the method described in Patent Document 1 has a problem that it cannot be applied during operation.

[0005] This disclosure has been made in view of the above circumstances, and an object thereof is to provide a control system, a control method, and a program capable of appropriately controlling devices during operation.

[0006] The control system according to this disclosure includes a plurality of information technology devices installed in a space, one or more air-cooling devices for air-cooling the space and at least one of the information technology devices, one or more liquid-cooling devices for liquid-cooling at least one of the information technology devices, an acquisition unit that acquires predetermined information related to the space at a predetermined cycle, a prediction unit that predicts a predetermined future state in the space based on the information using a prediction model that virtually reproduces the space on a computer, and a control unit that controls at least one of the information technology devices, the air-cooling devices, and the liquid-cooling devices based on the future state.

[0007] The control method relating to this disclosure includes the steps of acquiring predetermined information relating to a space, a plurality of information technology devices installed in the space, one or more air cooling devices for air cooling the space and at least one of the information technology devices, one or more liquid cooling devices for liquid cooling at least one of the information technology devices, and the space at predetermined intervals; predicting a predetermined future state in the space based on the information using a predictive model that virtually reproduces the space on a computer; and controlling at least one of the information technology devices, the air cooling devices, and the liquid cooling devices based on the future state.

[0008] The program relating to this disclosure causes a computer to perform the following steps: acquire predetermined information relating to a plurality of information technology devices installed in a space, one or more air cooling devices for air cooling the space and at least one of the information technology devices, one or more liquid cooling devices for liquid cooling at least one of the information technology devices, and the space at a predetermined interval; predict a predetermined future state in the space based on the information using a predictive model that virtually reproduces the space on a computer; and control at least one of the information technology devices, the air cooling devices, and the liquid cooling devices based on the future state.

[0009] According to the control system, control method, and program disclosed herein, equipment can be properly controlled during operation.

[0010] This is a block diagram showing an example configuration of a control system according to the embodiment of this disclosure. This is a diagram showing an example of information acquired by the acquisition unit according to the embodiment of this disclosure. This is a diagram showing an example of prediction results by the prediction unit according to the embodiment of this disclosure. This is a diagram showing an example of output of prediction results according to the embodiment of this disclosure. This is a diagram showing an example of machine learning of the machine learning model according to the embodiment of this disclosure. This is a diagram showing an example of items controlled by the control unit according to the embodiment of this disclosure. This is a diagram for explaining an example of operation of the control unit according to the embodiment of this disclosure. This is a diagram showing an example of output of the supply unit according to the embodiment of this disclosure. This is a diagram showing an example of output of the supply unit according to the embodiment of this disclosure. This is a diagram showing an example of output of the supply unit according to the embodiment of this disclosure. This is a flowchart showing an example of operation of a control system according to the embodiment of this disclosure. This is a schematic block diagram showing the configuration of a computer according to the embodiment of this disclosure.

[0011] Hereinafter, the control system, control method, and program according to the embodiments of this disclosure will be described with reference to Figures 1 to 14. In each figure, the same or corresponding components are given the same reference numerals, and their descriptions will be omitted as appropriate.

[0012] Figure 1 is a block diagram showing an example configuration of a control system according to an embodiment of the present disclosure, and shows an example configuration of a data center control system 100 including a control system 2 according to an embodiment of the present disclosure. In the example shown in Figure 1, the data center control system 100 comprises a data center 1, a control system 2, a DCIM (Data Center Infrastructure Management) 3, a user terminal 4, and a relay network NW. The data center 1, the control system 2, the DCIM 3, and the user terminal 4 can communicate with each other via the relay network NW.

[0013] Data Center 1 is a building for installing and operating IT equipment (information technology equipment) (or ICT (information and communication technology) equipment) such as servers and communication equipment. Data Center 1 comprises one or more server rooms 5. Server room 5 is an example of the configuration of the "space" related to this disclosure. One or more server racks 6, air-cooled equipment groups 7, liquid-cooled equipment groups 8, etc. are installed in Server Room 5. Hereafter, information technology equipment such as servers 61, air-cooled equipment included in air-cooled equipment group 7, liquid-cooled equipment included in liquid-cooled equipment group 8, etc. will also be simply referred to as "equipment".

[0014] The server rack 6 houses one or more information technology devices such as servers 61. As will be described later, the server rack 6 also includes some of the equipment within the air-cooled equipment group 7 and the liquid-cooled equipment group 8.

[0015] The air-cooling equipment group 7 includes multiple air-cooling devices such as one or more air conditioners 71 and one or more server fans 72. In this disclosure, "air-cooling equipment" refers to equipment that operates on electricity to cool equipment such as servers 61 or the server room 5 using air as a heat transfer medium. Examples of air-cooling equipment include air conditioners, fans, pumps, etc. The air conditioner 71 cools the air circulating within the server room 5. The server fan 72 is installed, for example, in a server 61 or server rack 6 and blows air onto the server 61.

[0016] The liquid cooling equipment group 8 includes multiple liquid cooling devices such as one or more rear door cooling devices 81, one or more direct chip cooling devices 82, and one or more immersion cooling devices 83. In this disclosure, "liquid cooling equipment" refers to powered equipment that uses liquid as a heat transfer medium to cool equipment such as a server 61. The rear door cooling device 81 forms the rear door of the server rack 6 and includes a refrigerant pipe 811 that cools the air heated by the heat dissipation from the server, and a rear door fan 812 that passes the heated air through. The direct chip cooling device 82 circulates refrigerant through a cold plate installed on the server chip (processor) mounted on the server 61 to dissipate heat from the server chip. The immersion cooling device 83 cools the server 61 and other equipment by immersing them in refrigerant in an immersion tank.

[0017] Note that multiple sensors are installed in server room 5 and inside each piece of equipment to measure the state of the room and each piece of equipment (or each piece of equipment has built-in sensors or measurement functions), but these are not shown in the diagram.

[0018] DCIM3 is a system (software) for managing the facilities and equipment within data center 1. It monitors and records the status of server room 5 and each piece of equipment within server room 5, and also controls the status of each piece of equipment.

[0019] User terminal 4 is, for example, a personal computer, tablet, or smartphone used by users of the various devices in the server room 5. In response to user operations, user terminal 4 monitors and controls the status of the various devices in the server room 5, either directly or via DCIM 3, etc. It also displays information provided by the control system 2 using a web browser, etc.

[0020] The control system (or control device) 2 is a component of the "control system" according to this disclosure and can be configured using one or more computers. The control system 2 comprises an acquisition unit 21, a prediction unit 22, a control unit 23, and a supply unit 24 as a functional block composed of a combination of hardware such as a computer and software such as a program executed by that computer.

[0021] The acquisition unit 21 repeatedly acquires predetermined information indicating the status of the server room 5 and each piece of equipment within the server room 5, for example, via DCIM 3, at a predetermined interval. Figure 2 shows an example of the information acquired by the acquisition unit 21. In the example shown in Figure 2, the classification of the cooling method, the equipment, the items of information acquired in real time (at a predetermined interval) for each piece of equipment, and the use of the information in the control system 2 are shown in correspondence. In the example shown in Figure 2, the acquisition unit 21 acquires the load factor of each server 61, the arrangement of each server 61, the airflow rate of each server fan 72, the outlet temperature of each air conditioner, the airflow rate of each rear door fan 812, the set (refrigerant) temperature of the refrigerant piping 811 of each rear door, the set (refrigerant) temperature of the cold plate of each direct chip cooling equipment 82, the set (refrigerant) temperature of the immersion tank of each immersion cooling equipment, etc., from DCIM 3 at a predetermined interval as setting conditions (use A) for the fluid dynamics simulation described later. Furthermore, the acquisition unit 21 acquires multiple temperatures and power consumption values ​​from DCIM 3 at predetermined intervals, as shown in Figure 2, for the purpose of verifying the consistency of the results of the fluid dynamics simulation described later (Purpose B). In addition, the acquisition unit 21 acquires the set temperature of the air conditioner 71 from DCIM 3 at predetermined intervals, as the purpose of verifying the consistency of the control by the control unit 23 described later (Purpose C). Note that each item shown in Figure 2 is an example and will differ depending on the installation status of the equipment in the server room 5.

[0022] The information shown in Figure 2 acquires the load factor, temperature, and location of each server 61 in all cases of cooling by air conditioning (air cooling), rear door cooling, direct chip cooling, and liquid immersion cooling. The load factor and temperature are obtained from the server 61, and the location (position of the server rack 6 and position within the server rack 6 (number of shelves)) is obtained from a location information sensor attached to the server 61. In addition, the temperature of each part of the air-cooled and liquid-cooled equipment and the power consumption of the components are also acquired. In the air-cooled method, the airflow rate discharged by the server fan 72, the set temperature of the air conditioner 71, and the outlet temperature are acquired. The airflow rate is output from the server 61, and the set temperature and outlet temperature are output from the air conditioner 71. In addition, temperature is acquired from temperature sensors attached to the cold aisle (the cooled space in the server room 5) and the hot aisle (the space on the server heat exhaust side in the server room 5). In the rear door cooling method, predetermined values ​​are obtained from the server 61 and refrigerant piping 811, similar to the air-cooled method. In addition, the fan airflow rate and set (refrigerant) temperature output from the rear door are acquired. In direct tip cooling and immersion cooling systems, the set (refrigerant) temperature output from each device is acquired. Each piece of information is associated with the identification information of that device.

[0023] Furthermore, the acquisition unit 21 acquires, for example, 3D information regarding the server room 5 and the installation of each piece of equipment within the server room 5, separate from the real-time information shown in Figure 2, via DCIM 3. This 3D information only needs to be acquired once, unless it changes. The 3D information includes, for example, the shape and size of the server room 5 above and below the floor, the shape, size, orientation, and position of each server rack 6, the shape, size, and position of each piece of equipment installed outside the server rack 6, including the air-cooling equipment group 7, the direction of airflow for each piece of equipment, and information representing the shape, size, and position of each piece of equipment installed outside the server rack 6, including the liquid-cooling equipment group 8. The acquisition unit 21 also acquires, for example, information indicating the performance of each piece of equipment, and information indicating the relationship between the operating state and power consumption of each piece of equipment, via DCIM 3. This information also only needs to be acquired once, unless it changes.

[0024] As described above, in this embodiment, the acquisition unit 21 acquires predetermined information at predetermined intervals relating to a plurality of servers 61 (information technology equipment) installed in the server room 5, one or more air-cooling devices including a group of air-cooling devices 7 for air-cooling the server room 5 and at least one server 61, one or more liquid-cooling devices for liquid-cooling at least one server 61, and the server room 5.

[0025] The prediction unit 22 predicts a predetermined future state within the server room 5 based on information acquired by the acquisition unit 21, using a prediction model 221 that virtually reproduces the server room 5 (space) on a computer. In this embodiment, the prediction unit 22 predicts the future state using either or both of a computational fluid dynamics model 2211 and a machine learning model 2212. In this embodiment, the state in the future state is, for example, the state of the fluid, and can include temperature at one or more points (locations) in the server room 5, flow velocity at one or more points, flow rate at one or more points, pressure at one or more points, temperature distribution at multiple points, flow velocity distribution at multiple points, flow rate distribution at multiple points, pressure distribution at multiple points, and multiple of these. However, the state targeted by this embodiment is the general state of physical quantities that can be predicted by the model, and is not limited to these examples.

[0026] The computational fluid dynamics model 2211 is a model that predicts the flow of liquids and gases by modeling the flow of fluids (liquids and gases) using equations based on computational fluid dynamics (CFD) and solving the equations using a computer. In this embodiment, the computational fluid dynamics model 2211 virtually reproduces each piece of equipment in a three-dimensional space and is constructed to output data representing the results of predicting the future temperature distribution and flow velocity distribution in the server room 5 by inputting the data shown as application A in Figure 2, for example. As shown in Figure 3, in the thermal fluid simulation using the computational fluid dynamics model 2211, the arrangement of data center equipment, server heat generation, airflow rate, air conditioning outlet temperature, airflow rate, liquid refrigerant temperature, etc. are input as conditions and the temperature of the servers and the space in a state that can be considered steady state is evaluated. Figure 3 shows an example of the prediction results using the computational fluid dynamics model 2211.

[0027] Furthermore, the prediction unit 22 may generate information for checking the load and temperature for each server 61, as shown in Figure 4, based on the temperature distribution prediction results using the computational fluid dynamics model 2211 and the information acquired by the acquisition unit 21. Figure 4 shows a plan view of the server racks 6 in the server room 5 along with the occupancy rate of the server racks 6, and an example of a table displayed when a server rack 6 is selected on the plan view. The user can check predetermined information using the user terminal 4 on a screen such as the one shown in Figure 4. In the example shown in Figure 4, servers "××" and "△△" are identification information (or names) based on the server 61 arrangement information. The CPU (processor) load corresponds to the load rate shown in Figure 2. The chip temperature is based on the current temperature distribution predicted by the prediction unit 22 (or information acquired from the server 61).

[0028] Furthermore, the machine learning model 2212 is a machine learning model (surrogate model) that uses the input data to the computational fluid dynamics model 2211 and the output data representing future states calculated using the computational fluid dynamics model 2211 as training data, and is trained to take the input data to the computational fluid dynamics model 2211 as input and output output data similar to the output data from the computational fluid dynamics model 2211. Figure 5 shows an example of the training of the machine learning model 2212. As shown in Figure 5, the machine learning model 2212 takes power consumption (load factor), air conditioner set temperature, liquid cooling equipment set temperature, computer node (location), etc. as explanatory variables and is trained to output the server room temperature distribution and flow velocity distribution in the future (for example, XX minutes ahead (where "XX" is an arbitrary number)) as the objective variables. A surrogate model is a model that acts as a substitute for (or replaces) a numerical simulation model, and generally, it can obtain output in a shorter time compared to performing a numerical simulation. Therefore, after training is complete (after the model is built), the machine learning model 2212 can be used to perform more predictions in the same amount of time as the computational fluid dynamics model 2211. Note that the machine learning model 2212 may be a model that uses a neural network, or it may not.

[0029] Furthermore, when operating each device, for example at the start of operation, the prediction unit 22 uses the computational fluid dynamics model 2211 to predict the future state and records the input and output data of the computational fluid dynamics model 2211. When a predetermined amount of input and output data has been accumulated, the prediction unit 22 constructs a machine learning model 2212 using the recorded input and output data. After constructing the machine learning model 2212, the prediction unit 22 can start predicting the future state using the machine learning model 2212. Note that the state in which the machine learning model 2212 has been constructed means that the machine learning model 2212 has been properly trained (for example, when a predetermined evaluation index related to machine learning exceeds (or falls below) a predetermined value).

[0030] The control unit 23 comprises a first control unit 231 and a second control unit 232. The first control unit 231 controls each device (at least one device) in the server room 5 by generating control signals (also called control parameters) based on predetermined rules derived from the prediction results of the prediction unit 22. The control unit 23 controls each device by transmitting control signals to each device directly or via DCIM 3, etc. The predetermined rules include, for example, a rule that if the temperature of a certain server 61 exceeds a predetermined temperature, the set temperature of a certain air conditioner 71 should be lowered by XX degrees. Alternatively, the predetermined rules include, for example, a rule that if the temperature of all servers 61 falls below a predetermined temperature, the set temperature of all air conditioners 71 should be raised by XX degrees.

[0031] Figure 6 shows an example of items controlled by the control unit 23 (control units 231 and 232). In the example shown in Figure 6, the items to be controlled include the load factor of each server 61 (however, the load factor is controlled only if there is a server 61 whose temperature exceeds a predetermined level even after controlling other items), the airflow of each server fan 72, the set temperature of each air conditioner 71, the set temperature of the refrigerant piping 811, the airflow of each rear door fan 812, the set (refrigerant) temperature of each cold plate, the set (refrigerant) temperature of each immersion tank, etc.

[0032] The second control unit 232, for example, uses a machine learning model 2212 to predict the temperature distribution and flow velocity distribution multiple times after a predetermined time while changing the setting conditions, satisfies the constraints regarding the temperature of the server 61, and controls each device (at least one device) by generating a control signal to control the load rate of the server 61, the set temperature for air cooling, the set temperature for liquid cooling, and the airflow rate for air cooling, so that the value of the indicator approaches a predetermined value (for example, to maximize or minimize the indicator). An example of an indicator related to energy efficiency is PUE (Power Usage Effectiveness). PUE is the value obtained by dividing the total power consumption of each device by the power consumption of the IT equipment, and the closer it is to 1, the higher the efficiency. However, the indicator related to energy efficiency may be other indicators. In addition, the indicator for operating costs may be, for example, electricity costs, predicted maintenance costs and replacement costs of equipment. In addition, other indicators such as the temperature of the processor may be used.

[0033] Furthermore, when the second control unit 232 determines the control parameters (Figure 6), it may generate a response surface that shows the sensitivity of the control parameters to indicators such as PUE, electricity cost, and CPU temperature, and determine the control parameters so that each indicator satisfies predetermined conditions. Figure 7 shows an example of control parameter determination by the second control unit 232. In the example shown in Figure 7, when the control parameters are the set temperature of air conditioner (1) and the set temperature of air conditioner (2), the set temperature of air conditioner (1) and the set temperature of air conditioner (2) are determined using PUE as an indicator, while satisfying constraint conditions regarding temperature (for example, the temperature of the most critical part). Surface RS1 is the response surface of PUE when the set temperature of air conditioner (1) and the set temperature of air conditioner (2) are changed. Surface RS2 is the response surface of temperature when the set temperature of air conditioner (1) and the set temperature of air conditioner (2) are changed. Surface TS1, shown by the dashed line, represents the temperature limit value. In the example shown in Figure 7, the optimal combination of the set temperatures of air conditioner (1) and air conditioner (2) is determined as the one that results in the lowest PUE within each range of the set temperatures of air conditioner (1) and air conditioner (2) where surface RS2 is less than or equal to surface TS1.

[0034] The method for setting control parameters in the control unit 23 (first control unit 231 and second control unit 232) is not limited to the above example. For example, control parameters may be determined using a machine learning model that has been trained based on data obtained from operational experience.

[0035] Furthermore, the providing unit 24 generates comparable information at at least two different points on the time axis for at least one of the following: information acquired by the acquisition unit 21 (current set temperature, power consumption, PUE, electricity cost calculated based on these, etc.), prediction results by the prediction unit 22 (current and future CPU temperature, etc.), and control results by the control unit 23 (future set temperature, power consumption, PUE, electricity cost, etc.), and provides this information to, for example, the user terminal 4. Figures 8 to 12 show examples of the information provided. All of these are examples of information comparing the current value with a value XX minutes later. Figure 8 shows an example of information comparing the CPU temperatures of multiple servers 61. Figure 9 shows an example of information comparing the set temperatures of each device. Figure 10 shows an example of information comparing the power consumption of each device. Figure 11 shows an example of information comparing PUE. Figure 12 shows an example of information comparing electricity costs.

[0036] Next, an example of the operation of the control system 1 will be described with reference to Figure 13. The process shown in Figure 13 is executed at predetermined intervals after the server room 5 starts operation. When the process shown in Figure 13 starts, first, the acquisition unit 21 acquires predetermined information from DCIM 3 (step S1). Next, the prediction unit 22 determines whether or not the machine learning model 2212 has been constructed (step S2). If the machine learning model 2212 has not been constructed (step S2: NO), the prediction unit 22 uses the computational fluid dynamics model 2211 to predict the temperature distribution and flow velocity distribution after a predetermined time (step S3). Next, the prediction unit 22 records the input data and output data (temperature distribution and flow velocity distribution after a predetermined time) of the computational fluid dynamics model 2211 (step S4). Next, the prediction unit 22 uses all the recorded input data and output data to train the machine learning model 2212 (step S5). Next, the first control unit 231 generates a control signal (step S6). Next, the control unit 23 outputs a control signal (step S7) and terminates the process shown in Figure 13.

[0037] On the other hand, if the machine learning model 2212 has already been constructed (step S2: YES), the prediction unit 22 uses the machine learning model 2212 to predict the temperature distribution and flow velocity distribution multiple times after a predetermined time while changing the setting conditions, and the second control unit 232 generates a control signal to control at least one of the server 61's load rate, set temperature for air cooling, set temperature for liquid cooling, and airflow rate for air cooling, so that the value of the indicator approaches a predetermined value (maximize or minimize the indicator), satisfying the constraints regarding the temperature of the server 61 and using at least one of energy efficiency and operating cost as indicators (step S8). Next, the control unit 23 outputs a control signal (step S7) and terminates the process shown in Figure 13.

[0038] (Effects) The control system, control method, and program according to this embodiment acquire predetermined information at predetermined intervals regarding a plurality of information technology devices installed in the server room 5 (space), one or more air cooling devices for air cooling the server room 5 and at least one information technology device, one or more liquid cooling devices for liquid cooling at least one information technology device, and the server room 5. Furthermore, a predetermined future state within the server room 5 is predicted based on the information acquired using a prediction model 221 that virtually reproduces the server room 5 on a computer. Based on the future state, at least one of the information technology devices, air cooling devices, and liquid cooling devices is controlled. With this configuration, each device can be controlled based on the results of predicting the future state. Therefore, according to the control system, control method, and program according to this embodiment, the devices can be appropriately controlled during operation.

[0039] Furthermore, the prediction unit 22 uses at least one of the following as the prediction model 221: a computational fluid dynamics model 2211 that calculates the temperature distribution and flow velocity distribution in the server room 5 as future states, and a machine learning model 2212 that has been trained to take input data as input data and output data as training data, using the input data to the computational fluid dynamics model 2211 and the output data representing the calculated future states as training data. In addition, when operating information technology equipment, air-cooled equipment and liquid-cooled equipment, the prediction unit 22 uses the computational fluid dynamics model 2211 to predict future states and records the input data and output data. After constructing the machine learning model 2212 using the recorded input data and output data, it starts predicting future states using the machine learning model 2212. With these configurations, for example, if the configuration or location of the equipment changes, the computational fluid dynamics model 2211 can be used to predict future states, and after sufficient data has been accumulated for training, the machine learning model 2212 can be used to predict future states.

[0040] Further, the control unit 23 controls at least one of the information technology device, the air-cooling device, and the liquid-cooling device by changing at least one of the load factor of the information technology device, the set temperature in air cooling, the set temperature / air volume in liquid cooling, and the air volume in air cooling so that the index approaches a predetermined value, satisfying at least the constraint conditions regarding the temperature of the information technology device and using at least one of energy efficiency and operation cost as an index. According to this configuration, the value of the control signal for each device can be made more appropriate.

[0041] Further, the providing unit 24 provides comparable provided information at at least two different points on the time axis for at least one of the information acquired by the acquiring unit 21, the prediction result by the predicting unit 22, and the control result by the control unit 23. According to this configuration, for example, a user can easily confirm the control content and the like.

[0042] Further, according to the present embodiment, for example, a prediction model under predetermined conditions is generated, and the air conditioner can be operated within a range that conforms to the energy-saving conditions. Therefore, it can be operated more efficiently compared to the case of controlling according to the situation. That is, there is a great advantage compared to the control of cooling after getting hot.

[0043] (Other Embodiments) Although the embodiments of the present disclosure have been described in detail with reference to the drawings above, the specific configuration is not limited to this embodiment, and design changes and the like within the scope not departing from the gist of the present disclosure are also included.

[0044] (Computer Configuration) FIG. 14 is a schematic block diagram showing the configuration of a computer according to the present embodiment. The computer 90 includes a processor 91, a main memory 92, a storage 93, and an interface 94. The above-described control system 2 is implemented in the computer 90. The operations of the above-described respective processing units are stored in the storage 93 in the form of a program. The processor 91 reads the program from the storage 93, expands it in the main memory 92, and executes the above processing according to the program. Further, the processor 91 secures a storage area corresponding to each of the above-described storage units in the main memory 92 according to the program.

[0045] The program may be for implementing a part of the functions to be performed by the computer 90. For example, the program may perform functions in combination with other programs already stored in storage, or in combination with other programs implemented in other devices. In other embodiments, the computer may be equipped with a custom LSI (Large Scale Integrated Circuit) such as a PLD (Programmable Logic Device) in addition to or instead of the above configuration. Examples of PLDs include PAL (Programmable Array Logic), GAL (Generic Array Logic), CPLD (Complex Programmable Logic Device), FPGA (Field Programmable Gate Array), etc. In this case, some or all of the functions implemented by the processor may be implemented by the integrated circuit.

[0046] Examples of storage 93 include HDDs (Hard Disk Drives), SSDs (Solid State Drives), magnetic disks, magneto-optical disks, CD-ROMs (Compact Disc Read Only Memory), DVD-ROMs (Digital Versatile Disc Read Only Memory), and semiconductor memory. Storage 93 may be an internal medium directly connected to the bus of the computer 90, or an external medium connected to the computer 90 via an interface 94 or a communication line. Furthermore, if this program is distributed to the computer 90 via a communication line, the computer 90 that receives the distribution may expand the program into the main memory 92 and execute the above processing. In at least one embodiment, storage 93 is a tangible storage medium that is not temporary.

[0047] <Note> The control system 2 described in each embodiment can be understood, for example, as follows.

[0048] (1) The control system 2 according to the first aspect includes a plurality of information technology devices installed in a space, one or more air cooling devices for air-cooling the space and at least one of the information technology devices, one or more liquid cooling devices for liquid-cooling at least one of the information technology devices, an acquisition unit 21 that acquires predetermined information related to the space at a predetermined cycle, a prediction unit 22 that predicts a predetermined future state in the space based on the information using a prediction model 221 that virtually reproduces the space on a computer, and a control unit 23 that controls at least one of the information technology devices, the air cooling devices, and the liquid cooling devices based on the future state. According to this aspect and the following aspects, the devices can be appropriately controlled during operation.

[0049] (2) The control system 2 according to the second aspect is the control system 2 in (1), wherein the prediction unit 22 uses at least one of a numerical fluid dynamics model 2211 that calculates a temperature distribution and a flow velocity distribution in the space as the future state, and a machine learning model 2212 that is machine-learned to input the input data and output the output data using the input data to the numerical fluid dynamics model and the output data representing the calculated future state as learning data, as the prediction model.

[0050] (3) The control system 2 according to the third aspect is the control system 2 in (1) or (2), wherein the prediction unit 22 predicts the future state using the numerical fluid dynamics model during operation of the information technology devices, the air cooling devices, and the liquid cooling devices, records the input data and the output data, constructs the machine learning model using the recorded input data and output data, and then starts predicting the future state using the machine learning model.

[0051] (4) The control system 2 according to the fourth embodiment is the control system 2 according to (1) to (3), wherein the control unit controls at least one of the information technology equipment, the air-cooled equipment and the liquid-cooled equipment by changing at least one of the load factor of the information technology equipment, the set temperature for air cooling, the set temperature for liquid cooling, and the airflow rate for air cooling, so that the indicator approaches a predetermined value, satisfying at least the temperature constraints of the information technology equipment and using at least one of energy efficiency and operating cost as an indicator.

[0052] (5) The control system 2 according to the fifth embodiment is the control system 2 according to (1) to (4), further comprising a providing unit 24 that provides comparable information at at least two different points on the time axis with respect to at least one of the information acquired by the acquisition unit, the prediction result by the prediction unit, and the control result by the control unit.

[0053] According to the control system, control method, and program disclosed herein, equipment can be properly controlled during operation.

[0054] 100...Data center control system 1...Data center 2...Control system 5...Server room 21...Acquisition unit 22...Prediction unit 221...Prediction model 2211...Computational fluid dynamics model 2212...Machine learning model 23...Control unit 24...Provision unit 61...Server

Claims

1. A control system comprising: an acquisition unit that acquires predetermined information relating to a plurality of information technology devices installed in a space, one or more air-cooling devices for air-cooling the space and at least one of the information technology devices, one or more liquid-cooling devices for liquid-cooling at least one of the information technology devices, and the space at predetermined intervals; a prediction unit that predicts a predetermined future state in the space based on the information using a prediction model that virtually reproduces the space on a computer; and a control unit that controls at least one of the information technology devices, the air-cooling devices, and the liquid-cooling devices based on the future state.

2. The control system according to claim 1, wherein the prediction unit uses as the prediction model at least one of a computational fluid dynamics model that calculates the temperature distribution and flow velocity distribution in the space as the future state, and a machine learning model that is trained to take the input data as input and output the output data as input data, using the input data to the computational fluid dynamics model and the output data representing the calculated future state as training data.

3. The control system according to claim 2, wherein the prediction unit predicts the future state using the computational fluid dynamics model when operating the information technology equipment, the air-cooled equipment and the liquid-cooled equipment, records the input data and the output data, constructs the machine learning model using the recorded input data and the output data, and then starts predicting the future state using the machine learning model.

4. The control system according to claim 3, wherein the control unit controls at least one of the information technology equipment, the air-cooled equipment, and the liquid-cooled equipment by changing the load factor of the information technology equipment, the set temperature for air cooling, the set temperature for liquid cooling, and the airflow rate for air cooling, so that the control unit satisfies at least the temperature constraints of the information technology equipment and at least one of energy efficiency and operating cost as indicators, so that the indicator approaches a predetermined value.

5. The control system according to any one of claims 1 to 4, further comprising a providing unit that provides comparable information at at least two different points on the time axis with respect to at least one of the information acquired by the acquisition unit, the prediction result by the prediction unit, and the control result by the control unit.

6. A control method comprising the steps of: acquiring predetermined information relating to a plurality of information technology devices installed in a space, one or more air cooling devices for air cooling the space and at least one of the information technology devices, one or more liquid cooling devices for liquid cooling at least one of the information technology devices, and the space itself at predetermined intervals; predicting a predetermined future state in the space based on the information using a predictive model that virtually reproduces the space on a computer; and controlling at least one of the information technology devices, the air cooling devices, and the liquid cooling devices based on the future state.

7. A program that causes a computer to perform the following steps: acquire predetermined information relating to a plurality of information technology devices installed in a space, one or more air cooling devices for air cooling the space and at least one of the information technology devices, one or more liquid cooling devices for liquid cooling at least one of the information technology devices, and the space itself, at predetermined intervals; predict a predetermined future state in the space based on the information using a predictive model that virtually reproduces the space on a computer; and control at least one of the information technology devices, the air cooling devices, and the liquid cooling devices based on the future state.

Images