Power amount control device and power amount control program

Abstract

A power amount control device (10) comprises: a room-by-room learning history information generation unit (120) that generates learning history information for each room; an air-conditioning power / temperature function approximation unit (130) that approximates an air-conditioning power function and a temperature function on the basis of the learning history information; and a power cost efficiency optimal solution calculation unit (151) that, when calculating the power consumption cost efficiency, executes pruning processing in which, at the end of each turn, only one pattern of the air-conditioning control level having the lowest total electricity charge up to the turn for each temperature level is held, and in which the initial temperature level of the next turn corresponding to the temperature at the end of the turn and the total electricity charge solution up to the turn are indicated.

Description

Electric energy control device and electric energy control program

[0001] The present invention relates to an electric energy control device and an electric energy control program for controlling the amount of power consumed by servers and air conditioners in each room in a data center (hereinafter sometimes referred to as "DC").

[0002] With the goal of reducing greenhouse gas emissions, the spread of facilities, equipment, and technologies to expand the use of renewable energy obtained from solar power, wind power, etc. is being promoted. However, since the supply of renewable energy is affected by changes in the natural environment and it cannot be stored, it is necessary to maximize the efficiency of renewable energy use by increasing demand in line with peak power supply. As a result, it is expected that electricity prices will tend to fluctuate throughout the day, for example, by making daytime electricity, when more electricity can be expected from solar power, cheaper than nighttime electricity.

[0003] In addition, the load within a DC also fluctuates throughout the day, and the system that controls the DC must take into account both the load arriving at the DC and fluctuations in electricity rates. Furthermore, the amount of data processing (load) in a DC is increasing year by year, and this also necessitates improving the power consumption efficiency of the DC as a whole (the amount of power consumed by the entire DC for a certain amount of data processing). In a DC, in addition to the power consumption of servers, the power consumption of air conditioning accounts for a large proportion, and therefore there is a need to reduce the power consumption of the DC as a whole.

[0004] Non-Patent Document 1 discloses a technology for optimizing the overall power consumption of a data center by taking into account the power consumption of air conditioners and servers (IT equipment). The air-conditioning-linked IT load placement optimization method for data centers described in Non-Patent Document 1 predicts future load trends on IT equipment by collecting operation information and monitoring information from the data center's IT equipment, and calculates the power increase of air conditioning equipment in response to the increase in power consumption of the IT equipment. Then, an optimization problem is solved to minimize the objective function, which is the power consumption of the data center, so that the load concentration rate on the IT equipment increases over time, i.e., the number of operating IT devices is reduced. This calculates the placement of IT loads (virtual machines) on the IT equipment that minimizes the power consumption of the data center.

[0005] Jun Okitsu and 4 others, "IT Load Allocation Optimization Method Linked with Air Conditioning for Eco-Friendly Data Centers," FIT (Forum on Information Technology) 2010, 9th Information Science and Technology Forum, RC-009

[0006] However, in the technology described in Non-Patent Document 1, a general-purpose rule-based standard that is not dependent on the equipment conditions that differ for each data center is adopted in the air conditioning power model used to calculate the power consumption of the air conditioning equipment. Therefore, it was difficult to perform optimization to reduce the power consumption of the data center while taking into account individual equipment conditions such as the location of the air conditioning equipment, airflow, server layout configuration within the data center, and thermal cooling efficiency.

[0007] Furthermore, no consideration was given to optimizing power cost efficiency by taking into account daily fluctuations in electricity rates, fluctuations in the load flowing into the DC, load allocation within the DC, and air conditioning control. Factors that affect the power consumption of air conditioning cooling for a given amount of load arriving at the DC include the air conditioning control level of the air conditioner (control parameter <1>) as well as the load allocation within the DC. Load allocation within the DC includes how much load is allocated to each room in the DC (load allocation pattern between DC rooms: control parameter <2>) and how the load allocated to each room is further allocated to the servers within that room (load allocation pattern within the DC room: control parameter <3>).

[0008] Here, we consider a "room-specific load allocation amount / load allocation pattern" (hereinafter referred to as "allocation allocation pattern") in which the predetermined time interval for updating the air conditioning control and the load allocation in each room is set to one turn (e.g., 30 minutes), the load allocation amount to each room is changed for the total DC load amount in each turn, and the load allocation amount to each room is further changed by the allocation ratio for each area within the room ("allocation control area 30" in Figure 6 described later).

[0009] After fixing one of these distribution patterns, that is, when the load distribution amount (load distribution pattern) of each room in each turn and the load distribution pattern in the DC room are fixed, the power consumption cost efficiency (E) that maximizes the air conditioning power charge efficiency in the DC total and the air conditioning control pattern for each room at that time are determined. Here, if the number of control stages of air conditioning control is "G", the number of rooms is "R", and the number of turns is "tn", the number of patterns of air conditioning control levels is G tn・R This G tn・R From the patterns, the pattern that minimizes the power consumption cost efficiency (E) is determined.

[0010] However, when calculating the above-mentioned power consumption cost efficiency (E) and the air conditioning control pattern for each room at that time, there was a problem that as the number of turns increased, the calculation processing time increased explosively, and the calculation resources such as memory required and the time required became enormous.

[0011] The present invention has been made in light of these points, and its objective is to control servers and air conditioners while taking into account fluctuations in the amount of load flowing into a DC and fluctuations in electricity rates, thereby increasing the power consumption cost efficiency of the entire DC and reducing the calculation time for the optimal solution that maximizes power consumption cost efficiency.

[0012] The power amount control device according to the present invention is a power amount control device that is communicatively connected to an in-room control device for each room that controls servers and air conditioners in the rooms in a data center, and that allocates the load amount to be processed in each of the rooms, wherein a plurality of servers are allocated in each placement control area that indicates a concentrated area for processing the load, and one or more air conditioners are allocated in each of the rooms, and the power amount control device controls the servers and the air conditioners in a turn that indicates a predetermined control time for a learning pattern that combines a plurality of stages of initial temperature in the room, a plurality of stages of server load to be loaded on the entire room, a plurality of load placement patterns that indicate the ratio at which the server load is distributed and allocated to the placement control areas, and a plurality of stages of air conditioning control levels of the air conditioners, thereby acquiring the air conditioning power consumption of the entire room and the temperature in the room at the end of the turn from the in-room control device for each room, as a learning history for each of the learning patterns, and generates learning history information for each of the rooms, and the air conditioning power consumption in a room is defined as an air conditioning power function, and for each of the load placement patterns and the air conditioning control levels, the air conditioning power function is approximated based on the learning history information, and the temperature in the room at the end of the turn for the load amount allocated to the room is defined as a temperature function, and for each of the load placement patterns and the air conditioning control levels, an air conditioning power / temperature function approximation unit approximates the temperature function based on the learning history information, and for each of the rooms, the air conditioning power / temperature function approximation unit approximates the temperature function based on the learning history information, and defines the total sum of the electricity charges based on the air conditioning power consumption in each of the successive turns and the total electricity charges based on the server load in each of the successive turns divided by the total load amount in each of the successive turns as the power consumption cost efficiency, and calculates the power consumption cost efficiency of the entire data center when the air conditioning control level for each of the rooms is changed using the approximated air conditioning power function and the temperature function, while keeping the load allocation pattern for allocating the total load to the data center to each of the rooms in each successive turn and the load placement pattern in the rooms fixed;and a power cost efficiency optimal solution calculation unit that calculates, as an optimal solution, the air conditioning control level for each room that maximizes the power consumption cost efficiency in the entire data center, wherein the power cost efficiency optimal solution calculation unit performs pruning processing that, when calculating the power consumption cost efficiency, retains, at the end of each turn, only one pattern of air conditioning control level that has the lowest total electricity charge up to that turn for each temperature level that is a plurality of stages of room temperature that can be achieved at the end of that turn, and sets that as the initial temperature level for the next turn that corresponds to the temperature at the end of that turn and the total electricity charge solution up to that turn.

[0013] According to the present invention, it is possible to control servers and air conditioners taking into account fluctuations in the load flowing into the DC and fluctuations in electricity rates, thereby increasing the power consumption cost efficiency of the entire DC and reducing the calculation time for the optimal solution that maximizes power consumption cost efficiency.

[0014] 1 is a diagram showing the difference in air conditioning power consumption when a server load is applied to two rooms in a DC and the air conditioners are controlled (setting temperature and air volume). FIG. 2 is a diagram showing allocation patterns in which the load distribution rate is changed for each allocation control area. FIG. 3 is a diagram showing the air conditioning power consumption when the load amount for each allocation pattern is loaded on servers arranged in each allocation control area, measured in normal operation and degenerate operation. FIG. 4 is a diagram showing the overall configuration of a power amount control system including a power amount control device according to the present embodiment. FIG. 5 is a diagram for explaining that a room according to the present embodiment is a cohesive logical area in which a group of air conditioners has a cooling effect on a group of servers. FIG. 6 is a diagram showing the configuration within one room of a DC according to the present embodiment. FIG. 7 is a functional block diagram showing an example of the configuration of an in-room control device according to the present embodiment. FIG. 8 is a diagram showing an example of the control stages of the air conditioning control level according to the present embodiment. FIG. 9 is a diagram for explaining a learning pattern for acquiring a learning history according to the present embodiment. FIG. 10 is a functional block diagram showing an example of the configuration of a power amount control device according to the present embodiment. FIG. 11 is a diagram illustrating an air conditioning power function according to the present embodiment. FIG. 12 is a diagram illustrating an example of a temperature function according to the present embodiment. FIG. 13 is a diagram for explaining allocation allocation patterns according to the present embodiment. FIG. 14 is a diagram for explaining a pruning method for power consumption cost efficiency calculation according to the present embodiment. 1 is a diagram for explaining a pruning method for calculating power consumption cost efficiency according to the present embodiment. FIG. 2 is a diagram showing a calculation formula for power consumption cost efficiency according to <Method 1> when the initial temperature deviates. FIG. 3 is a diagram showing a calculation formula for power consumption cost efficiency according to <Method 2> when the initial temperature deviates. FIG. 4 is a sequence diagram showing the flow of processing executed by a power amount control system including a power amount control device according to the present embodiment. FIG. 5 is a flowchart showing the flow of processing executed by a power cost efficiency optimal solution update unit of the power amount control device according to the present embodiment. FIG. 6 is a diagram showing a simulation result of the calculation time for power consumption cost efficiency when the pruning method according to the present embodiment is used. FIG. 7 is a hardware configuration diagram showing an example of a computer that realizes the functions of the power amount control device and the in-room control device according to the present embodiment. FIG. 8 is a diagram showing the simulation result of the calculation time for power consumption cost efficiency when the pruning method according to the present embodiment is not used.

[0015] <Problem and Outline> First, the problem of the present invention will be described in detail, and then the outline of the present invention will be described.

[0016] The power consumption cost efficiency (E) of the entire DC is calculated for N turns after changing the air conditioning control level pattern for each room in each distribution arrangement pattern. Specifically, when calculating this power consumption cost efficiency (E), one pattern from each load arrangement pattern is sequentially fixed (selected) and the power consumption cost efficiency (E) is calculated. This power consumption cost efficiency (E) is defined as the sum of the total electricity charges based on the air conditioning power consumption for each successive turn and the total electricity charges based on the server load for each successive turn, divided by the total load for each successive turn, as shown in the following formula (1):

[0017]

[0018] Here, the coefficients and parameters for the jth turn (tj turn) are as follows: "m" is the number of rooms, and "n" is the number of turns. "c[tj]" indicates the power consumption cost efficiency coefficient for the turn in question, specifically the electricity rate coefficient. For example, it is (7 yen / kWh), and is the coefficient for converting the amount of power consumed into the electricity rate. "k" is the coefficient for converting the server load into server power consumption (server load / power coefficient). The parameters in formula (1) are as follows:

[0019]

[0020] Furthermore, the total load flowing into DC1000 from turn "1" to turn "n", which is indicated by X[tj] in the lower part of equation (1), is given by the following equation (2).

[0021]

[0022] In calculating this power consumption cost efficiency (E), the number of patterns for the air conditioning control level is G, where G is the number of control stages, R is the number of rooms, and tn is the number of turns. tn・R It becomes a street.

[0023] As a concrete example of calculating power consumption cost efficiency (E), the following figure shows the calculation processing time on a laptop PC for a scenario with 7 to 13 turns for one room, a fixed allocation pattern, and full air conditioning control level set to "Level 6." The optimal solution was calculated for the next 6 hours (12 turns), with each turn being 30 minutes. As shown in Figure 22, the calculation processing time increases exponentially as the number of turns increases, resulting in enormous amounts of computational resources, such as memory, and the required time. In this simulation, based on the number of processing patterns and measured time when the number of turns was 7, the predicted calculation time was 4 minutes when the number of turns was 7, but when the number of turns was 12 (6 hours), the predicted calculation time was approximately 518 hours (21.6 days).

[0024] The power consumption control device according to this embodiment performs the following process when calculating the overall room power consumption cost efficiency (E) by changing the air-conditioning control level for each turn, from the first turn to the Nth turn, and adding up the air-conditioning power cost for each turn. At the end of each turn, the power consumption control device retains as a branch (selected data) only one air-conditioning control level pattern that offers the lowest total electricity cost for each temperature level among the room temperature levels available at the end of the turn. The power consumption control device then uses the initial temperature for the next turn and the total electricity cost solution up to the previous turn to prune (not select) all branches (data other than the selected data) with the second lowest or lower electricity cost for the same temperature level, thereby reducing the number of patterns used in calculating the optimal solution for power efficiency (details will be described later). This method of retaining and calculating only one air-conditioning control level pattern with the lowest total electricity cost for each temperature level as a branch (selected data) is hereinafter referred to as the "pruning method" (pruning process). By implementing this "pruning method," the power consumption control device can significantly reduce the time required to calculate the optimal solution.

[0025] Next, an embodiment of the present invention (hereinafter referred to as "this embodiment") will be described. First, as a premise, load allocation, which is a factor affecting the power consumption of air conditioning cooling for a certain amount of load arriving in a DC, will be specifically described.

[0026] Figure 1 shows the difference in air conditioning power consumption within one room when a load (server load) is applied to two rooms ("r1" and "r2") within a DC and the air conditioner is controlled (setting the set temperature and air volume).

[0027] Here, as indicated by the symbol α, when comparing "No. r1-2" in room "r1" with "No. r2-3" in room "r2," the server load is the same, "60 kW," the room temperature before measurement began was also roughly the same, and the air conditioner's set temperature (blowout temperature) and airflow settings are the same. Nevertheless, when comparing the air conditioning power consumption within one room, there is a difference: room "r1" consumes 7.50 (kW) while room "r2" consumes 6.06 (kW). Furthermore, as indicated by the symbol β, when comparing "No. r1-3" in room "r1" with "No. r2-1" in room "r2," the server load is the same, "20 kW," the room temperature before measurement began was also roughly the same, and the air conditioner's set temperature (blowout temperature) and airflow settings are the same. Nevertheless, when comparing the air conditioning power consumption within each room, room "r1" consumes 1.70 (kW), while room "r2" consumes 5.70 (kW).

[0028] In other words, even if the same air conditioner settings (set temperature, air volume) are used for each room with the same distribution load, the amount of power consumed by air conditioning cooling will differ. Therefore, for the same load, it is better to distribute the load to the room with the lowest air conditioning power consumption, and it is necessary to optimize the load between rooms for a fixed load flowing into the DC.

[0029] Next, we will explain the air conditioning power consumption depending on the load allocation pattern within a data center. Assume that a single room is divided into areas where multiple servers are allocated, and these are controlled as "allocation control zones" (server groups). Figure 2 shows an example in which "Zone 1" through "Zone 4" are established as allocation control zones (see Figure 6 below). The patterns for allocating the load allocated to this room are, for example, allocation patterns "1" through "6." For example, allocation pattern "1" allocates 100% of the load to "Zone 1," 0% to "Zone 2," 0% to "Zone 3," and 0% to "Zone 4." The other allocation patterns "2" through "6" similarly allocate the load to "Zone 1" through "Zone 4" according to the load distribution ratios shown in Figure 2.

[0030] Figure 3 shows the air conditioning power consumption measured in both normal and degraded operation when the load for each placement pattern is applied to servers located in each placement control area. In normal operation, air conditioning control is performed at 100% of the air conditioning capacity, and the air conditioning power consumption is roughly the same for placement patterns "1" to "6." On the other hand, when degraded control (optimal control that minimizes air conditioner output within a range that satisfies temperature rewards based on set temperatures, etc.) is used, the air conditioning power consumption differs for each placement pattern. In Figure 3, placement pattern "3" allows for the greatest reduction in air conditioning power consumption (see symbol γ in Figure 3).

[0031] In other words, when optimally controlling air conditioners, if a certain load amount allocated to a room is allocated to each allocation control area using different allocation patterns, the amount of air conditioning power consumption will differ. Therefore, for the same load amount, it is better to allocate the load to the allocation control area using an allocation pattern that consumes less air conditioning power, and it is necessary to optimize the allocation pattern for the certain load amount allocated to the room.

[0032] As described above, maximizing power consumption cost efficiency when the load flowing into the data center and the electricity rate fluctuate during a certain consecutive time period requires coordinated optimization of the air conditioning control level of each room (control parameter 1), the load distribution amount between rooms (control parameter 2), and the load allocation pattern within the rooms (control parameter 3). A power consumption control system including a power consumption control device according to this embodiment can improve the power consumption cost efficiency of the entire data center by taking into account the parameters of the air conditioning control level of each room, the load distribution amount between rooms, and the load allocation pattern within the rooms. Furthermore, by using a pruning technique, described in detail below, the power consumption control device can significantly reduce the time required to calculate the optimal solution for power consumption cost efficiency and air conditioning control level. This is explained in detail below.

[0033] <Electricity Amount Control System> FIG. 4 is a diagram showing the overall configuration of an electric energy control system 1 including an electric energy control device 10 according to this embodiment.

[0034] 4, the power amount control system 1 is configured to include a DC (data center) 1000 having a plurality of rooms 500 (rooms "A," "B," and "C" in FIG. 4) each equipped with a plurality of servers 3 and one or more air conditioners 4, an in-room control device 20 provided for each room 500 and communicatively connected to the plurality of servers 3 and one or more air conditioners 4 in the room 500, and a power amount control device 10 communicatively connected to each in-room control device 20 and controlling the amount of power consumed by the entire DC 1000. The power amount control device 10 and each in-room control device 20 may be provided inside the DC 1000 or may be provided in a location separate from the DC 1000.

[0035] In this embodiment, the room 500 is defined as a logical area in which a specific group of air conditioners exerts a cooling effect on a group of servers. Normally, the room 500 in the DC 1000 can be considered a single area surrounded by walls, but in the DC 1000, a group of servers may be located in a single, very large space. In such cases, depending on the location of the air conditioners and the layout and configuration of the servers in the room 500, servers may not be affected by the cooling effect of the air conditioners 4 even if they are physically located in the same room 500.

[0036] Therefore, for example, as shown by reference numeral 51 in FIG. 5 , if air conditioner group "A" and air conditioner group "B" are located separately even within a single physical room, before executing control within room 500, area <1> is identified and set as the room by controlling the air conditioning of only air conditioner group "A" to lower the temperature by a predetermined amount (for example, 2° C.). Similarly, area <2> is identified and set as the room by controlling the air conditioning of only air conditioner group "B" to lower the temperature by a predetermined amount (for example, 2° C.). Note that, as shown by reference numeral 52 in FIG. 5 , if air conditioner group "A" affects all servers and air conditioner group "B" affects all servers, area <1> and area <2> are set as the same room.

[0037] 4 , in the DC 1000, an in-room control device 20 is provided in association with each room 500 defined as a logical area as described above. The in-room control device 20 corresponding to each room 500 may acquire status information (such as the amount of power consumed by air conditioning) of the air conditioners 4 provided in the room 500 via an air conditioning management device (not shown) or transmit air conditioning control information (such as set temperature and air volume), or may be directly connected to and communicate with each air conditioner 4 without going through the air conditioning management device.

[0038] The in-room control device 20 is also connected for communication with each server 3 in the room 500. The in-room control device 20 may be connected for communication via a server management device (not shown) to acquire status information (server load, etc.) of the servers 3 provided in the room 500 and transmit control information (load allocation, etc.), or may be connected for communication directly with the servers 3.

[0039] It should be noted that the DC1000 will be described as one in which a virtualization platform is constructed and operated. Known open source virtualization platforms include OpenStack (registered trademark), which is software for constructing cloud environments, and Kubernetes (registered trademark), which is software for operating and managing containerized workloads and services. OpenStack is primarily used for managing and operating physical machines and virtual machines (VMs). Kubernetes is primarily used for managing and operating containers. In this specification, an application virtualized in a virtualization platform (composed of one or more containers, one or more VMs, etc.) is referred to as a virtual resource. In Kubernetes, the smallest execution unit of an application is a pod composed of one or more containers.

[0040] In the DC 1000 of this embodiment, the servers 3 accommodated are divided into areas where multiple servers 3 (server groups) are arranged, as shown in Figure 6, and are controlled as "placement control areas." This placement control area 30 is an area that accommodates a group of servers on which virtual resources are arranged, and indicates a group of areas that handle load. Figure 6 shows an example in which placement control areas "1" to "4" are provided.

[0041] In this embodiment, "air-conditioning controlled areas" are provided in association with the placement-controlled areas 30 of the server groups, as shown in FIG. 6 . The air-conditioning controlled areas 40 are areas in which the room temperature effect of air-conditioning control is measured, and are assumed to face either the intake or exhaust side of the servers 3. Air blown from the air conditioners 4 is blown out from the air-conditioning controlled areas 40 on the intake side (air-conditioning controlled areas "1," "2," "5," and "6" in FIG. 6 ) via piping installed, for example, under the floor of the DC 1000. Air whose temperature has risen due to the heat from each server 3 is taken in from the suction ports of the piping installed in the air-conditioning controlled areas 40 on the exhaust side (air-conditioning controlled areas "3" and "4" in FIG. 6 ), generating an airflow that returns to the air conditioners 4.

[0042] A plurality of sensors (temperature sensors, etc.) are installed in each of the air-conditioning controlled areas 40. Furthermore, in each placement controlled area 30, temperature sensors are installed at the intake and exhaust ports of each server 3 or of a server 3 selected within the area. Information obtained from these sensors (sensor information) can be acquired by the in-room control device 20 or power amount control device 10 associated with the room via a communication line, etc.

[0043] In the power amount control system 1 including the power amount control device 10 according to this embodiment, the power amount control device 10 acquires air-conditioning cooling learning data (room temperature and air-conditioning power amount) obtained by changing the initial temperature, load (server load), load allocation pattern, and air-conditioning control level in each room from the room-inside control device 20 and stores the data as learning history information. The power amount control device 10 then approximates the "air-conditioning power function" and "temperature function" for each room based on the learning history information (details will be described later). The power amount control device 10 generates a distribution allocation pattern (details will be described later) for each room in each turn based on input information on the total load amount in the DC over N turns. For each distribution allocation pattern, the power amount control device 10 then calculates the power cost efficiency of the entire DC over N turns when the air-conditioning control level in each room is changed, and extracts the air-conditioning control level for each room that provides the highest power cost efficiency of the entire DC for that pattern. In this case, at the end of each turn, the power amount control device 10 retains as a branch (selected data) only one pattern of air-conditioning control levels that results in the lowest total electricity cost up to that turn for each temperature level among the room temperature levels possible at the end of the turn, and uses this to determine the initial temperature for the next turn and the total electricity cost solution up to the previous turn. After using this pruning method, the power amount control device 10 determines, from all load distribution patterns, a distribution arrangement pattern and air-conditioning control levels for each room in that pattern that maximizes the power consumption cost efficiency of the entire DC. As a result, the power amount control system 1 equipped with the power amount control device 10 can improve the power consumption cost efficiency of the entire DC and reduce the calculation time for the optimal solution that maximizes power consumption cost efficiency.

[0044] Next, the power amount control device 10 and the in-room control device 20 that constitute the power amount control system 1 according to this embodiment will be specifically described.

[0045] <Room Control Device> First, the in-room control device 20 will be described. The in-room control device 20 is provided in association with each room 500 and has the same configuration. FIG. 7 is a functional block diagram showing an example of the configuration of the in-room control device 20 according to this embodiment. Upon receiving a learning history collection instruction from the power amount control device 10, the in-room control device 20 changes the initial temperature, load (server load allocated to the room 500), load allocation pattern, and air-conditioning control level, controls the server 3 and air conditioner 4, collects predetermined air-conditioning cooling learning data (room temperature, air-conditioning power consumption, etc.) as learning history, and transmits it to the power amount control device 10. The in-room control device 20 also controls the server 3 and air conditioner 4 in the room 500 according to the load and load allocation pattern based on the allocation allocation pattern determined by the power amount control device 10 to maximize the power consumption cost efficiency, and according to the determined air-conditioning control level.

[0046] The in-room control device 20 is configured by a computer including a control unit 200 , an input / output unit 250 , and a storage unit 260 .

[0047] The input / output unit 250 inputs and outputs information between the power amount control device 10 and each device (each server 3 and each air conditioner 4) in the room 500. The input / output unit 250 is composed of a communication interface that transmits and receives information via a communication line, and an input / output interface that inputs and outputs information between an input device such as a keyboard and an output device such as a monitor, which are not shown.

[0048] The storage unit 260 is configured with a hard disk, flash memory, RAM (Random Access Memory), etc. The storage unit 260 temporarily stores programs for executing the functions of the control unit 200 and information necessary for processing by the control unit 200.

[0049] The control unit 200 is responsible for all the processing executed by the in-room control device 20, and as shown in Figure 7, is composed of a temperature measurement unit 210, a learning history collection unit 220, an in-room server control unit 230, and an in-room air conditioning control unit 240.

[0050] The temperature measurement unit 210 measures the temperature inside the room 500 using a temperature sensor provided inside the room 500. The temperature measurement unit 210 may also obtain information on the outside air temperature using a temperature sensor provided outside the DC 1000.

[0051] The following temperatures are set in advance as the room temperature, for example: - The current temperature of a predetermined fixed-point sensor (temperature sensor) in the room 500 - The overall average of the sensor temperatures in each air-conditioning controlled zone 40 (FIG. 6) in the room 500 - The average temperature of each air-conditioning controlled zone 40 on the air intake side or the air-conditioning controlled zone 40 on the air exhaust side in the room 500 The following temperature ranges are defined for the room temperature: As a temperature range, the average value (one value) of multiple temperatures is divided into multiple ranges of values. For example, the temperature T is divided into temperature ranges of "18°C to 23°C," "24°C to 29°C," and "30°C to 35°C."

[0052] The temperature measurement unit 210 measures a predetermined room temperature as the "initial room temperature" when learning history collection starts. The temperature measurement unit 210 also measures the room temperature at the end of collection of learning history for a predetermined time (predetermined turn) as the "room temperature at the end of a turn." If the outside air temperature is set as a parameter of the learning history information, the temperature measurement unit 210 acquires information from a temperature sensor installed outside the DC 1000 and outputs the outside air temperature information to the learning history collection unit 220.

[0053] When the learning history collection unit 220 receives learning history collection instruction information from the power amount control device 10, it collects information on the initial room temperature, server load (server load allocated to the room), load allocation pattern, air conditioning power consumption per turn when controlled based on a learning pattern in which the air conditioning control level has been changed, the room temperature at the end of the turn, and the room temperature reward pass / fail judgment. Note that a "turn" is a predetermined interval related to the control of the server 3 and air conditioner 4 in the DC 1000. Specifically, one turn is defined as the predetermined time interval for updating the air conditioning control and the load allocation in the room 500. For example, as described above, one turn is set to 30 minutes.

[0054] Here, with regard to the air conditioning control of the air conditioner 4, the air conditioning control output value, which indicates the air conditioning cooling capacity, is divided into multiple stages according to intensity, and these are defined as air conditioning control levels. For example, FIG. 8 shows an example in which a unified air conditioning control level (set temperature and air volume) is set across one or more air conditioners 4. As shown in FIG. 8, air conditioning control level "1" is a set temperature of 32°C and an air volume of 24 Hz. Air conditioning control level "2" is a set temperature of 28°C and an air volume of 30 Hz. The higher the intensity of the air conditioning control level, up to air conditioning control level "5," the higher the cooling capacity. However, air conditioning power consumption tends to increase.

[0055] FIG. 9 shows an example in which control stages (set temperature and air volume) of the air conditioning control level are set for each of multiple air conditioners 4. As shown in FIG. 9, for example, at air conditioning control level "1," control stages (set temperature and air volume) are set for "Air Conditioning 1 (Air Conditioner "1")" and "Air Conditioning 2 (Air Conditioner "2")." Note that in FIG. 9, the same control stage is set for all air conditioners 4 at the same air conditioning control level, but parameters may be changed for each air conditioner 4 to achieve finer control level division. Furthermore, the control stages of the air conditioning control level are not limited to set temperature and air volume, and may include, for example, air direction, etc.

[0056] The learning history collection unit 220 executes control of a total of l x m x n x o patterns (hereinafter referred to as "predetermined learning patterns") in its own room 500 by changing the initial room temperature (l), server load (m), server load placement pattern (n), and air conditioning control level (o), and acquires information on the amount of air conditioning power consumed at that time, the room temperature at the end of the turn, and the room temperature reward pass / fail judgment as learning history.

[0057] 10, the learning history collection unit 220 sets the initial temperature (initial temperature inside the room) to three levels: "18°C," "24°C," and "30°C," and sets the server load to three levels: "30kW," "60kW," and "90kW." Regarding the load allocation pattern, the example shown in FIG. 10 indicates that the server load for the room 500 is allocated as "25%-25%-25%-25%, "50%-50%-0%-0%, and "0%-0%-50%-50%," respectively, for the four allocation control zones "1," "2," "3," and "4" (see FIG. 6). In the example shown in FIG. 10, air conditioning control is set in three stages: "Set temperature 32°C: Air volume 24Hz" (air conditioning control level "1" in FIG. 8), "Set temperature 24°C: Air volume 36Hz" (air conditioning control level "3" in FIG. 8), and "Set temperature 16°C: Air volume 48Hz" (air conditioning control level "5" in FIG. 8).

[0058] The learning history collection unit 220 acquires information on the air conditioning power consumption for a turn as a result of execution of a predetermined learning pattern from the air conditioning power measurement unit 242 (described later). The learning history collection unit 220 also acquires information on the room temperature at the end of the turn from the temperature measurement unit 210. The room temperature reward pass / fail judgment is information indicating the result of the judgment (pass / fail) on whether the temperature after the turn exceeded a predetermined room temperature threshold (room temperature threshold). The room temperature threshold is a threshold (limit temperature) set by the system management side in each room 500 of the DC 1000, and is set on the assumption that control will not be performed when the temperature threshold is exceeded. For patterns that "fail" the room temperature reward pass / fail judgment, since they will not be executed in actual operation, the learning history collection unit 220 may not transmit the data to the power amount control device 10. This reduces the calculation load on the power amount control device 10.

[0059] The learning history collecting unit 220 transmits the learning history executed in a predetermined learning pattern, together with the identification information of the in-room control unit 20 itself, to the power amount control device 10. Note that information on each parameter set in this predetermined learning pattern may be registered in advance in the learning history collecting unit 220, or may be acquired from a management device (not shown) of the DC 1000 or the power amount control device 10.

[0060] When generating this learning history of the predetermined pattern, the learning history collection unit 220 instructs the in-room server control unit 230 and the in-room air conditioning control unit 240 to allocate loads in the room 500 and to control air conditioning.

[0061] The in-room server control unit 230 (FIG. 7) causes the server 3 to execute control based on the load placement pattern for the placement control area 30 (FIG. 6) set within the room 500 when collecting the learning history (learning stage) and during operation (operation stage), and measures the server power consumption for each turn. This in-room server control unit 230 includes a load placement pattern setting unit 231 and a server power measurement unit 232.

[0062] The load placement pattern setting unit 231 sets a placement pattern in which the server load amounts set by the learning history collecting unit 220 are placed in each placement control area 30 ( FIG. 6 ) in the room 500. Then, the load placement pattern setting unit 231 generates and places virtual resources (VMs, containers, etc.) in each placement control area 30 according to the set placement pattern, and executes the load processing set when the learning history was collected. Furthermore, during operation (operation stage), the load placement pattern setting unit 231 sets, for the server groups in each placement control area 30 ( FIG. 6 ) in the room 500, a placement pattern indicated by a pattern that maximizes the power consumption cost efficiency of the entire DC, as determined by the power amount control device 10.

[0063] The server power measurement unit 232 measures the server power consumption (server power consumption) when each server 3 executes a process in the allocation pattern set by the load allocation pattern setting unit 231 .

[0064] The room air-conditioning control unit 240 controls the air conditioner 4 to perform air-conditioning control at a preset control stage (set temperature and air volume) of the air-conditioning control level, and measures the air-conditioning power consumption. This room air-conditioning control unit 240 includes an air-conditioning control execution unit 241 and an air-conditioning power measurement unit 242.

[0065] The air conditioning control execution unit 241 controls each air conditioner 4 in the room 500 at the air conditioning control level set by the learning history collection unit 220. Furthermore, during operation (operation stage), the air conditioning control execution unit 241 executes air conditioning control of the air conditioners 4 in each room 500 at the air conditioning control level in a pattern determined by the power amount control device 10 that maximizes the power consumption cost efficiency of the entire DC.

[0066] The air conditioning power measurement unit 242 measures the amount of air conditioning power consumed by the air conditioner 4 in the room 500 for each turn when collecting the learning history and during operation.

[0067] <Electricity Amount Control Device> Next, the electric energy control device 10 according to this embodiment will be described. FIG. 11 is a functional block diagram showing an example configuration of the electric energy control device 10 according to this embodiment. The electric energy control device 10 collects learning histories from each of the in-room control devices 20 corresponding to each room 500 and generates learning history information 550 for each room 500. The electric energy control device 10 then approximates the "air-conditioning power function" and "temperature function" for each room based on the learning histories. The electric energy control device 10 generates a distribution arrangement pattern for each room in each turn based on input information on the total load amount within the DC over N turns. For each distribution arrangement pattern, the electric energy control device 10 then calculates the power cost efficiency of the entire DC over N turns when the air-conditioning control level in each room is changed, and extracts the air-conditioning control level for each room that provides the highest power cost efficiency of the entire DC for the corresponding pattern. In this case, at the end of each turn, the power amount control device 10 retains as a branch (selected data) only one pattern of air-conditioning control levels that results in the lowest total electricity cost up to that turn for each possible room temperature level at the end of the turn, and uses this pattern to determine the initial temperature for the next turn and the total electricity cost solution up to the previous turn. Using this pruning method (pruning process), the power amount control device 10 determines, from all load distribution patterns, a distribution arrangement pattern and air-conditioning control levels for each room in that pattern that maximizes the power consumption cost efficiency of the entire DC. This allows the power amount control system 1 equipped with the power amount control device 10 to increase the power consumption cost efficiency of the entire DC and reduce the calculation time for the optimal solution that maximizes power consumption cost efficiency.

[0068] Furthermore, at the beginning of each turn during operation, the power amount control device 10 compares the deviation between the predicted value of the initial room temperature and the actual temperature, and if there is a deviation exceeding a predetermined value, it recalculates the power consumption cost efficiency for subsequent turns and updates the optimal solution. During this recalculation, the power amount control device 10 can also use a pruning method (pruning process) to reduce the calculation time for the optimal solution.

[0069] The power amount control device 10 is configured by a computer including a control unit 100 , an input / output unit 170 , and a storage unit 180 .

[0070] The input / output unit 170 inputs and outputs information between each in-room control device 20, an external system management device (not shown), etc. This input / output unit 170 is composed of a communication interface that transmits and receives information via a communication line, and an input / output interface that inputs and outputs information between an input device such as a keyboard and an output device such as a monitor, neither of which is shown.

[0071] The storage unit 180 is configured with a hard disk, flash memory, RAM (Random Access Memory), etc. This storage unit 180 temporarily stores programs for executing each function of the control unit 100 and information necessary for the processing of the control unit 100. For example, various coefficients such as the power consumption cost efficiency coefficient "c[tj]" (power consumption cost efficiency coefficient (electricity rate coefficient) at turn j) (i.e., information on the trend of the electricity rate coefficient) and the load / server power proportional coefficient "k" are stored in the storage unit 180, which will be described later. In addition, the storage unit 180 stores the total load amount within the DC 1000 for N turns (information on the load amount flowing into the DC in each turn) from a management device of the power amount control system 1 (FIG. 4), etc.

[0072] The control unit 100 is responsible for all the processing executed by the power amount control device 10, and as shown in FIG. 11, is composed of a room definition unit 110, a room-specific learning history information generation unit 120, an air conditioning power / temperature function approximation unit 130, an allocation arrangement pattern generation unit 140, an optimal control calculation unit 150 (a power cost efficiency optimal solution calculation unit 151, a control pattern determination unit 152), and a power cost efficiency optimal solution update unit 160.

[0073] The room definition unit 110 defines rooms 500 within the DC 1000 as consolidated logical areas where a specific group of air conditioners exerts a cooling effect on a group of servers. Specifically, when a predetermined trigger condition is met, for example, at a predetermined time interval (e.g., every three months or six months), when a new service process is started, or when maintenance or equipment upgrades are performed on the air conditioners 4, the room definition unit 110 transmits air conditioning control information to the room control device 20 to operate the group of air conditioners located in the consolidated area. As a result, the room definition unit 110 identifies servers 3 that are affected by the cooling effect of the group of air conditioners (e.g., a drop in the server intake temperature by 2°C or more) based on temperature information obtained from a temperature sensor, and sets the installation area of ​​the server 3 as the room 500.

[0074] The per-room learning history information generating unit 120 acquires the learning history executed in the above-described predetermined learning pattern from each of the in-room control devices 20 corresponding to each room 500. Then, the per-room learning history information generating unit 120 classifies the learning history executed in the predetermined learning pattern by initial temperature range, generates learning history information 550 for each room 500, and stores it in the storage unit 260.

[0075] For example, let us say that the "initial temperature range of 18°C ​​to 23°C" is "learning data set A." Let us say that the "initial temperature range of 24°C to 29°C" is "learning data set B." Let us say that the "initial temperature range of 30°C to 35°C" is "learning data set C." In this way, for each initial temperature range, the learning history information 550 stores information on the initial temperature, server load, load allocation pattern, and air conditioning control level as learning data sets, in association with the air conditioning power consumption at that time, the room temperature at the end of the turn, and the room temperature reward pass / fail judgment. If the outdoor temperature is included as a parameter of the predetermined learning pattern, the room-specific learning history information generation unit 120 may add the outdoor temperature classified into predetermined ranges to the learning history information 550. This predetermined learning pattern may be preset in the room-specific learning history information generation unit 120 or may be obtained from the management device (not shown) of the DC 1000.

[0076] The per-room learning history information generation unit 120 transmits, for example, learning history collection instruction information including information on a predetermined learning pattern to each in-room control device 20. As a result, the per-room learning history information generation unit 120 acquires the learning history in each room 500 and generates learning history information 550.

[0077] The air conditioning power / temperature function approximation unit 130 approximates the air conditioning power function and the temperature function for each load placement pattern and air conditioning control level (each combination of load placement pattern and air conditioning control level) for each room 500. The equations obtained by approximating the air conditioning power function and the temperature function are used when calculating the power cost efficiency, which will be described later.

[0078] The air conditioning power and temperature function approximation unit 130 defines the air conditioning power function for each room "ri" (i is a positive integer) as the amount of air conditioning power consumption relative to the load in the room 500. The air conditioning power and temperature function approximation unit 130 also defines the temperature function as the temperature at the end of a turn relative to the load in the room 500.

[0079] Here, the air conditioning control level at the jth (j is a positive integer) turn "tj" in room "ri" is set to "A[ri][tj]". The air conditioning control level at the turn "tj" in room "ri" is uniquely determined, for example, to "A(1)" (air conditioning control level "1" shown in Figure 8). The load placement pattern at the turn "tj" in room "ri" is set to "p[ri][tj]".

[0080] The air conditioning power / temperature function approximation unit 130 defines the air conditioning power consumption for the load (server load) assigned to the room 500 and approximates the air conditioning power function using the learning history information 550 for the room 500. Figure 12 shows an example of air conditioning control when the load placement pattern is "p[r1][t1]," the initial temperature is "19°C," and the air conditioning control level is "A(1)." Learning data for loads of "60 kW," "80 kW," and "100 kW" are obtained from learning history information (e.g., "learning data set A") for the "initial temperature range of 18°C ​​to 23°C," which includes the initial temperature of 19°C, to obtain the corresponding air conditioning power consumption. This allows the formula indicated by symbol a in Figure 12 to be obtained as an approximation of the air conditioning power function.

[0081] The air conditioning power / temperature function approximation unit 130 defines the temperature function as the temperature at the end of a turn for the load allocated to the room 500, and approximates the temperature function using the learning history information 550 for that room 500. Figure 13 shows an example of air conditioning control when the load placement pattern is "p[r1][t1]," the initial temperature is "19°C," and the air conditioning control level is "A(1)." Learning data for loads of "60 kW," "80 kW," and "100 kW" are obtained from learning history information (e.g., "learning data set A") for the "initial temperature range of 18°C ​​to 23°C," which includes the initial temperature of 19°C, to obtain the corresponding temperature at the end of a turn. This allows the equation indicated by symbol b in Figure 13 to be obtained as an approximation of the temperature function.

[0082] Returning to FIG. 11 , as described above, the distribution arrangement pattern generation unit 140 generates a “room-specific load distribution amount / load arrangement pattern” (hereinafter referred to as an “distribution arrangement pattern”) in which one turn is a predetermined time interval for updating the air conditioning control and the load arrangement within the room 500, the load distribution amount to each room 500 is changed for each turn with respect to the total load amount of DC1000, and the load distribution amount to each room 500 is further changed by the arrangement ratio for each area (arrangement control area 30) within the room 500.

[0083] FIG. 14 is a diagram illustrating a distribution arrangement pattern according to this embodiment. Here, the load amount flowing into DC1000 in each turn is "X[tj]". In this case, Σx[ri][tj] = X[tj] is the constraint. Note that "x[ri][tj]" is the load amount in room "ri" at turn "tj". Also, information on the total load amount in DC1000 for N turns is assumed to be obtained in advance as input information.

[0084] In the example shown in FIG. 14 , when "X[t1] = 50" (total load amount for turn "1": 50) is given, the distribution arrangement pattern generation unit 140 generates a load distribution pattern, for example, with the change amount in increments of 10. In this case, as the load distribution pattern (pt1) for turn "1," a pattern that distributes 0 kW to room "1" and 50 kW to room "2" (x[r1][t1] = 0, x[r2][t1] = 50), a pattern that distributes 10 kW to room "1" and 40 kW to room "2" (x[r1][t1] = 10, x[r2][t1] = 40), etc. are generated. Here, the load distribution pattern for turn "1" is defined as "F(t1)." Furthermore, when "X[t2] = 40" (total load for turn "2": 40) is given, the distribution arrangement pattern generation unit 140 generates, as the load distribution pattern (pt2) for turn "2", a pattern that distributes 0 kW to room "1" and 40 kW to room "2" (x[r1][t2] = 0, x[r2][t2] = 40), a pattern that distributes 10 kW to room "1" and 30 kW to room "2" (x[r1][t2] = 10, x[r2][t2] = 30), etc. Here, the load distribution pattern for turn "2" is defined as "F(t2)".

[0085] As shown in FIG. 14, the distribution placement pattern generation unit 140 generates "h" ("h": positive integer) patterns (three patterns in FIG. 14) such as "25%-25%-25%-25%", "50%-50%-0%-0%", "0%-0%-50%-50%" and the like for each area (placement control area 30) as load distribution patterns to be distributed to the areas (placement control areas 30) in the room 500. In this case, the number of distribution placement patterns in turn "1" (t1) is "F(t1)×h 2". The number of allocation patterns in turn "2" (t2) is "F(t2) × h 2 Therefore, when the load flowing into the DC 1000 from turn "1" to turn "n" (t1 to tn) is distributed to each of R rooms, the distribution arrangement pattern generation unit 140 calculates a total of "F(t1)×F(t2)×...×F(tn)×h R・F(tn) A distribution arrangement pattern of the " pattern is generated.

[0086] The distribution arrangement pattern generation unit 140 may acquire information on the load distribution amount of each room 500 with respect to the total load amount of the DC 1000 for each turn, and information on the allocation ratio at which the load distribution amount to each room 500 is allocated to each area (allocation control zone 30) within the room 500 from an external system management device, etc. In this case, the distribution arrangement pattern for each turn is uniquely determined.

[0087] Returning to FIG. 11 , the optimal control calculation unit 150 calculates the power consumption cost efficiency (E) of the entire DC 1000 over N turns by changing the air-conditioning control level pattern for each distribution arrangement pattern. At the end of each turn, the optimal control calculation unit 150 retains as a branch (selected data) only one air-conditioning control level pattern that provides the lowest total electricity cost up to that turn for each possible room temperature level at the end of the turn, and uses this data to determine the initial temperature for the next turn and the total electricity cost solution up to the previous turn. The optimal control calculation unit 150 then calculates the distribution arrangement pattern and air-conditioning control level for each room 500 that maximizes the power consumption cost efficiency (E) for the entire DC 1000 as the optimal solution. The optimal control calculation unit 150 includes a power cost efficiency optimal solution calculation unit 151 and a control pattern determination unit 152.

[0088] The power cost efficiency optimal solution calculation unit 151 calculates the power consumption cost efficiency (E) of the entire DC 1000 over N turns when the air-conditioning control level pattern is changed for each room 500 in each allocation arrangement pattern. At the end of each turn, the power cost efficiency optimal solution calculation unit 151 retains as a branch (selected data) only one air-conditioning control level pattern that provides the lowest total electricity cost up to that turn among the room temperature levels possible at the end of the turn, and uses this as the initial temperature for the next turn and the total electricity cost solution up to the previous turn. When calculating the power consumption cost efficiency (E) of the entire DC 1000, the power cost efficiency optimal solution calculation unit 151 does not include data from the learning data set stored in the learning history information 550 that has failed the room temperature reward pass / fail assessment. This also enables the power cost efficiency optimal solution calculation unit 151 to speed up the calculation of the power consumption cost efficiency (E).

[0089] When calculating this power consumption cost efficiency (E), the power cost efficiency optimal solution calculation unit 151 fixes (selects) one of the generated distribution arrangement patterns. By doing so, it is possible to uniquely determine the distribution load amount and load arrangement pattern for each room in each turn. Then, the power cost efficiency optimal solution calculation unit 151 calculates the power consumption cost efficiency (E) when the air conditioning control level pattern is changed for each room 500 in the fixed (selected) load arrangement pattern. The power cost efficiency optimal solution calculation unit 151 fixes (selects) each load arrangement pattern in turn and calculates the power consumption cost efficiency (E).

[0090] Specifically, the power cost efficiency optimal solution calculation unit 151 calculates the power consumption cost efficiency (E) for each air-conditioning control level pattern using the following formula (3): The power consumption cost efficiency (E) is defined as the sum of the total electricity charges based on the air-conditioning power consumption for each successive turn and the total electricity charges based on the server load for each successive turn, divided by the total load for each successive turn.

[0091]

[0092] Here, the coefficients and parameters for the jth turn (tj turn) are as follows. Note that the same symbols are used for parameters similar to those in Equation (1). "m" is the number of rooms, and "n" is the number of turns. "c[tj]" indicates the power consumption cost efficiency coefficient for the turn in question, specifically, the electricity rate coefficient. For example, it is (7 yen / kWh), and is a coefficient for converting the amount of power consumption into the electricity rate. "k" is a coefficient (server load / power coefficient) for converting the server load into server power consumption. This "k" is calculated based on the server load set for each server 3 and the server power (server power consumption) measured by the server power measurement unit 232 (Figure 7) when generating the learning history. Furthermore, the total load flowing into DC 1000 from turn "1" to turn "n," indicated by X[tj] in the lower part of Equation (3), is the same as in Equation (2) above. The parameters in Equation (3) are as follows:

[0093]

[0094] As shown above, "the air conditioning control level with the lowest total electricity cost up to the end of turn tj at the temperature level Tm[tj] at the end of turn tj": A'[ri][tj][Tm[tj]] is always one. The difference between formula (1) and formula (3) is that the air conditioning control level for the "initial temperature at the start" and "air conditioning power consumption" is A[ri][tj-1] in formula (1), but A'[ri][tj-1][Tm[tj-1]] in formula (3). In formula (1), branches were branched for each air conditioning control level at each turn. However, by using formula (3), the number of branches (selected data) carried over to the next turn at the end of each turn can be limited to M, the number of possible temperature levels.

[0095] 15A and 15B are diagrams illustrating the pruning method (pruning process) in this embodiment. To briefly and specifically explain the pruning method, an example of the pruning method will be described for a case where the number of rooms is "1," the temperature level is "3," and the air-conditioning control level is "6." The temperature levels are assumed to be "M1: (15.0°C or higher and lower than 21.0°C)," "M2: (21.0°C or higher and lower than 24.0°C)," and "M3: (24.0°C or higher and lower than 35.0°C)." Also, assume that a room load of "30" is applied in turn 1, a room load of "40" in turn 2, a room load of "80" in turn 3, and a room load of "80" in turn 4.

[0096] The power cost efficiency optimal solution calculation unit 151 executes a pruning method in which the branches (selected data) carried over to the next turn at the end of each turn are limited to a maximum of three, which is the number of possible initial temperature levels for the next turn. If the branches have the same temperature level at the end of the turn, the power cost efficiency optimal solution calculation unit 151 carries over only the branch with the lowest total electricity bill up to that point. This will be described in detail below with reference to Figures 15A and 15B.

[0097] First, in turn 1, as shown in FIG. 15A , the power cost efficiency optimal solution calculation unit 151 calculates the end temperature (temperature level) and the electricity rate for the case where the air conditioning control levels are C1 to C6 at that initial temperature level (here, “initial temperature level 1”). The power cost efficiency optimal solution calculation unit 151 calculates the temperature at the end of the turn based on the learning history information 550 for the corresponding initial temperature level and the temperature function calculated by the air conditioning power / temperature function approximation unit 130, and determines the temperature level from the temperature at the end of the turn. The power cost efficiency optimal solution calculation unit 151 also calculates the power consumption based on the load applied to that turn and the air conditioning power function calculated by the air conditioning power / temperature function approximation unit 130. The power cost efficiency optimal solution calculation unit 151 then calculates the electricity rate using the calculated power consumption and the power consumption cost efficiency coefficient (electricity rate coefficient) “c[tj]” for that turn.

[0098] The diagram indicated by the reference numeral 501 in FIG. 15A shows the calculation results at the end of turn 1. Here, for each temperature level at the end of the turn, the power cost efficiency optimal solution calculation unit 151 retains as a branch (selected data) only one pattern of the air conditioning control level with the lowest total electricity cost up to that turn. Specifically, in turn 1, the power cost efficiency optimal solution calculation unit 151 retains a branch for the air conditioning control level "C3" with the lowest electricity cost among the temperature level "M1," resulting in an electricity cost of 250 yen. Similarly, the power cost efficiency optimal solution calculation unit 151 retains a branch for the air conditioning control level "C2" with the lowest electricity cost among the temperature level "M2," resulting in an electricity cost of 200 yen. Similarly, the power cost efficiency optimal solution calculation unit 151 retains a branch for the air conditioning control level "C1" with the lowest electricity cost among the temperature level "M3," resulting in an electricity cost of 150 yen. The power cost efficiency optimal solution calculation unit 151 does not retain any branch other than the branch selected as the one with the lowest electricity rate for each temperature level, and does not use it for the calculation of the next turn.

[0099] Next, the power cost efficiency optimal solution calculation unit 151 calculates the temperatures (temperature levels) and electricity rates at the end of turn 2 for three initial temperature levels (M1 to M3) and six air conditioning control levels (C1 to C6). The calculation method is the same as for turn 1, but the power cost efficiency optimal solution calculation unit 151 calculates the total electricity rate for turn 1 and turn 2.

[0100] Then, as shown by reference numeral 502 in FIG. 15A , in turn 2, the power cost efficiency optimal solution calculation unit 151 retains the branch for the air conditioning control level "C2" (turn 1) → "C1" (turn 2) with the lowest electricity rate among the temperature level "M1", resulting in an electricity rate of 345 yen. Similarly, the power cost efficiency optimal solution calculation unit 151 retains the branch for the air conditioning control level "C2" → "C2" with the lowest electricity rate among the temperature level "M2", resulting in an electricity rate of 390 yen. Similarly, the power cost efficiency optimal solution calculation unit 151 retains the branch for the air conditioning control level "C3" → "C1" with the lowest electricity rate among the temperature level "M3", resulting in an electricity rate of 400 yen. The power cost efficiency optimal solution calculation unit 151 retains only the branch selected as the one with the lowest electricity rate among each temperature level, and does not use it in the calculation for the next turn. Furthermore, when selecting the branch with the lowest electricity rate from each temperature level, the power cost efficiency optimal solution calculation unit 151 does not include data for which the room temperature reward pass / fail judgment was "failed" (here, "28.0°C" or higher). Therefore, in turn 2, the calculation does not include branches where the temperatures at the end of the turn were 33.5°C, 28.6°C, or 31.2°C, as shown by the double strikethrough in the diagram of reference numeral 502 in Fig. 15A.

[0101] The power cost efficiency optimal solution calculation unit 151 repeats this pruning method from turn 3 indicated by reference numeral 503 in Figure 15B onwards, and obtains the following result in turn 4 indicated by reference numeral 504 in Figure 15B. The power cost efficiency optimal solution calculation unit 151 retains the branch with the air conditioning control level "C3" (turn 1) → "C1" (turn 2) → "C1" (turn 3) → "C1" (turn 4), which has the lowest electricity rate among the temperature level "M1", and an electricity rate of "725 yen". Similarly, the power cost efficiency optimal solution calculation unit 151 retains the branch with the air conditioning control level "C3" → "C1" → "C1" → "C3", which has the lowest electricity rate among the temperature level "M2", and an electricity rate of "795 yen". Similarly, the power cost efficiency optimal solution calculation unit 151 retains the branch with the air conditioning control level "C2" → "C2" → "C1" → "C1" which has the cheapest electricity rate among the temperature level "M3" and an electricity rate of "760 yen."Then, of the three retained branches, the power cost efficiency optimal solution calculation unit 151 selects the branch with the cheapest electricity rate of "725 yen," and calculates an optimal solution for controlling the air conditioning control level in this case from "C3" → "C1" → "C1" → "C1" in turns 1 to 4.

[0102] 11 , the control pattern determination unit 152 determines, from among all load allocation patterns calculated by the power cost efficiency optimal solution calculation unit 151, Eall = min(E{p}) that maximizes the power consumption cost efficiency (E) for the entire DC 1000. That is, the control pattern determination unit 152 further determines, as optimal solutions, the allocation allocation pattern x[ri][tj][p[ri][tj]] and the air-conditioning control level A[ri][tj] for each room 500 that maximizes the power consumption cost efficiency (E) for the entire DC 1000 from among the power consumption cost efficiencies (E) that maximize the cost efficiency calculated for each selected allocation allocation pattern. Then, the control pattern determination unit 152 sets the load allocation (load allocation pattern between rooms, load allocation pattern within a room) for each room 500 based on the optimal solutions for the determined allocation allocation pattern and air-conditioning control level, and executes air-conditioning control at the determined air-conditioning control level for each room 500.

[0103] At the beginning of each turn during operation, the power cost efficiency optimal solution update unit 160 collects initial temperature information from the temperature measurement unit 210 of each in-room control device 20 and compares the collected initial temperature (hereinafter referred to as the "actual measurement value") with the initial room temperature for the turn indicated by the room temperature at the end of the previous turn in the allocation arrangement pattern calculated as the optimal solution by the optimal control calculation unit 150 (hereinafter referred to as the "predicted value"). If there is a discrepancy between the actual measurement value and the predicted value exceeding a predetermined value, the power cost efficiency optimal solution update unit 160 recalculates the power consumption cost efficiency for the turn in which the discrepancy was found and thereafter, using the actual measurement value, and updates the optimal solution. In this case, the power cost efficiency optimal solution update unit 160 applies the same pruning method (pruning process) as that performed by the power cost efficiency optimal solution calculation unit 151 to the power efficiency calculations from the recalculation turn up to the Nth turn. That is, the power cost efficiency optimal solution update unit 160 substitutes the room initial temperature (actual measurement value) of the latest relevant turn into the calculation formula for power consumption cost efficiency (formula (3)), and updates the optimal solution.

[0104] There are two methods for recalculating the power consumption cost efficiency (E) during operation: a method (method 1) in which recalculation is performed only for the room 500 in which a temperature deviation has occurred, and a method (method 2) in which, if a temperature deviation has occurred in any of the rooms 500, the latest initial temperatures (actual measured values) of all the rooms 500 are substituted into the calculation formula for the power consumption cost efficiency (E) and recalculated.

[0105] <Method 1> In method 1, the power cost efficiency optimal solution update unit 160 substitutes the latest actual measurement value into equation (3) for power consumption cost efficiency (E) only for the room 500 where a temperature deviation has occurred, and updates the air conditioning control level for the following turn.

[0106] Specifically, if, at the beginning of a certain turn X of a certain room [ri], the room temperature function u[ri] (x[ri][tj-1]p[ri][tj-1], A'[ri][tj-1][Tm[tj-1]]) deviates by ±P°C or more from the actual measured temperature R[t(X-1)], the power cost efficiency optimal solution update unit 160 replaces the temperature of the room temperature function u[ri] (x[ri][tj-1]p[ri][tj-1], A'[ri][tj-1][Tm[tj-1]]) with the value of the actual measured temperature R[X-1], and updates the power consumption cost efficiency (E) from turn X to turn N.

[0107] The formula for calculating the power consumption cost efficiency (E) when there is a difference in the initial temperatures of rooms "1" and "2" among rooms "1", "2", and "3" during the X turn is formula (4) shown in Figure 16.

[0108] 16, the parameters enclosed by dotted lines are the actual measured values ​​to be re-substituted for rooms "1" and "2." The parameters enclosed by dashed lines are the air-conditioning control levels to be recalculated for rooms "1" and "2."

[0109] The power cost efficiency optimal solution update unit 160 recalculates the power consumption cost efficiency (E) from the X turn to the N turn using equation (4). As a result, the power cost efficiency optimal solution update unit 160 updates the remaining air conditioning control levels A[ri][tX], A[ri][tX+1], ..., A[ri][tN] from the X turn to the N turn for rooms "1" and "2" where a temperature discrepancy has occurred, so as to minimize the power consumption cost efficiency (E).

[0110] Here, the power cost efficiency optimal solution update unit 160 recalculates a maximum of (G x M x (n-X) + G) x Rk + M x Rk patterns of air conditioning control levels, where G is the number of control stages, M is the number of temperature levels, Rk is the number of rooms where a deviation has occurred, and X is the number of turns during recalculation, and determines the pattern that minimizes the power consumption cost efficiency (E).

[0111] <Method 2> In method 2, when a temperature deviation occurs in any of the rooms 500, the power cost efficiency optimal solution update unit 160 substitutes the latest actual measurement value into equation (3) for power consumption cost efficiency (E) for all the rooms 500, and updates the air conditioning control level for the following turn.

[0112] When a deviation of a predetermined value or more occurs between the initial temperature and the actual measured value in rooms "1" and "2" among rooms "1", "2", and "3" during the X turn, the power cost efficiency optimal solution update unit 160 calculates the power consumption cost efficiency (E) using formula (5) shown in FIG. 17 .

[0113] 17, the parameters enclosed by dotted lines are the actual measured values ​​to be re-assigned for rooms "1," "2," and "3." The parameters enclosed by dashed lines are the air conditioning control levels to be recalculated for rooms "1," "2," and "3."

[0114] The power cost efficiency optimal solution update unit 160 recalculates the power consumption cost efficiency (E) from the Xth turn to the Nth turn using equation (5). As a result, the power cost efficiency optimal solution update unit 160 re-updates the remaining air-conditioning control levels A[ri][tX], A[ri][tX+1], ..., A[ri][tN] from the xth turn to the Nth turn of all rooms 500 so as to maximize the power consumption cost efficiency (E).

[0115] Here, the power cost efficiency optimal solution update unit 160 recalculates up to (G x M x (n-X) + G) x R + M x R patterns of air conditioning control levels, where G is the number of control stages, M is the number of temperature levels, R is the total number of rooms, and X is the number of turns during recalculation, and determines the pattern that minimizes the power consumption cost efficiency (E).

[0116] The power cost efficiency optimal solution update unit 160 applies the updated air conditioning control level for each room 500 from the next turn after the calculation to update the power consumption cost efficiency (E) is completed. Even if a turn crossing occurs during the calculation to update the power consumption cost efficiency (E) and a condition for room temperature deviation occurs, the current calculation continues until the calculation is completed, and the update process does not have to be started again from the beginning. Furthermore, the power cost efficiency optimal solution update unit 160's process of updating the power consumption cost efficiency (E) can be applied in an environment where the calculation time required for the update is relatively short compared to the length of N turns.

[0117] <Processing Flow> Next, we will explain the processing flow executed by the power amount control system 1. Fig. 18 is a sequence diagram showing the processing flow executed by the power amount control system 1 including the power amount control device 10 according to this embodiment.

[0118] First, in step S1, when a predetermined trigger condition is met (for example, when a predetermined time interval has elapsed), the room definition unit 110 of the power amount control device 10 transmits air conditioning control information for operating a group of air conditioners arranged in a concentrated area to the room control device 20. As a result, the room definition unit 110 identifies the server 3 that will be affected by the cooling action of the group of air conditioners based on temperature information obtained from the temperature sensor, and sets the installation area of ​​the server 3 as a room 500.

[0119] Next, the per-room learning history information generating unit 120 of the power amount control device 10 transmits learning history collection instruction information to each in-room control device 20 (step S2). Note that this learning history collection instruction information may include information on a predetermined learning pattern (initial room temperature, server load, server load allocation pattern, air conditioning control level), or the predetermined learning pattern may be set in advance in the learning history collecting unit 220 of the in-room control device 20.

[0120] The learning history collection unit 220 of the in-room control device 20 sets the initial room temperature, server load, server load placement pattern, and air conditioning control level and executes control by instructing the in-room server control unit 230 and the in-room air conditioning control unit 240 to perform control using a predetermined learning pattern. As a result, the learning history collection unit 220 collects information on the air conditioning power consumption, the in-room temperature at the end of the turn, and the in-room temperature reward pass / fail judgment corresponding to the initial room temperature, server load, server load placement pattern, and air conditioning control level as learning history and transmits it to the power amount control device 10 (step S3).

[0121] The learning history collection unit 220 sets the initial temperature (initial room temperature) as a predetermined learning pattern, for example, to three levels: "18°C," "24°C," and "30°C," and the server load level to three levels: "30kW," "60kW," and "90kW." The learning history collection unit 220 also sets the server load level for the room 500 as a load allocation pattern, for example, to be "25%-25%-25%-25%, "50%-50%-0%-0%, and "0%-0%-50%-50%." The learning history collection unit 220 then acquires information on the initial room temperature and the room temperature at the end of the turn from the temperature measurement unit 210. The learning history collection unit 220 compares the information on the room temperature reward pass / fail judgment with the temperature threshold information stored in its own memory unit 260 ( FIG. 7 ), and determines whether the result exceeds the threshold as "fail" or whether the result is "pass" if the result is equal to or below the threshold.

[0122] Next, the per-room learning history information generating unit 120 of the power amount control device 10 acquires the learning history from each of the in-room control devices 20 in each room 500. Then, the per-room learning history information generating unit 120 classifies the learning history executed in the predetermined learning pattern by initial temperature range, generates learning history information 550 for each room 500 (step S4), and stores it in the storage unit 180 (FIG. 11).

[0123] Next, the air conditioning power / temperature function approximation unit 130 of the power amount control device 10 uses the learning history information 550 to approximate the air conditioning power function and temperature function for each load placement pattern and air conditioning control level (each combination of load placement pattern and air conditioning control level) for each room 500 (step S5).

[0124] Next, the distribution arrangement pattern generation unit 140 of the power amount control device 10 sets one turn as a predetermined time interval for updating the air conditioning control and the load arrangement within the room 500, changes the load distribution amount to each room 500 for the total load amount of DC1000 for each turn, and further changes the load distribution amount to each room 500 by the distribution ratio for each area (arrangement control zone 30) within the room 500 to generate a ``room-specific load distribution amount / load arrangement pattern'' (distribution arrangement pattern) (step S6).

[0125] The power cost efficiency optimal solution calculation unit 151 of the power energy control device 10 then calculates the power consumption cost efficiency (E) of the entire DC 1000 over N turns when a single distribution arrangement pattern is fixed and the air-conditioning control level pattern is changed for each room in each distribution arrangement pattern using a pruning method based on Equation (3), and determines the air-conditioning control level for each room 500 that maximizes the power consumption cost efficiency (E) (step S7). Specifically, at the end of each turn, the power cost efficiency optimal solution calculation unit 151 retains as a branch (selected data) only one air-conditioning control level pattern that minimizes the total electricity cost up to that turn for each possible room temperature level at the end of the turn, and uses this data to calculate the power consumption cost efficiency (E) by determining the initial temperature for the next turn and the total electricity cost solution up to the previous turn. The power cost efficiency optimal solution calculation unit 151 then determines the air-conditioning control level for each room 500 that maximizes the power consumption cost efficiency (E).

[0126] Next, the control pattern determination unit 152 of the power consumption control device 10 determines, from among the distribution arrangement patterns determined by the power cost efficiency optimal solution calculation unit 151, the load arrangement pattern that maximizes the power consumption cost efficiency (E) and the air conditioning control level for that load arrangement pattern as the optimal solution (step S8).

[0127] Next, the control pattern determination unit 152 transmits the load allocation (inter-room load allocation pattern, intra-room load allocation pattern) in each room 500 and the setting information of the determined air-conditioning control level for each room based on the determined optimal solution to each in-room control device 20 (step S9). As a result, each in-room control device 20 can execute control according to the inter-room load allocation pattern and intra-room load allocation pattern in each turn, and can also execute air-conditioning control of the air conditioners 4 at the determined air-conditioning control level in each turn.

[0128] <Power consumption cost efficiency update process> Next, a description will be given of the power consumption cost efficiency (E) update process executed by the power cost efficiency optimal solution update unit 160 of the power amount control device 10. Fig. 19 is a flowchart showing the flow of the process executed by the power cost efficiency optimal solution update unit 160 of the power amount control device 10 according to this embodiment.

[0129] First, at the beginning of each turn during operation, the power cost efficiency optimal solution update unit 160 acquires the actual temperature of each room 500 from each in-room control device 20. Then, the power cost efficiency optimal solution update unit 160 compares the actual temperature of each room 500 with the initial room temperature (predicted value) calculated as the optimal solution determined by the control pattern determination unit 152 to maximize the power consumption cost efficiency (E) for the entire DC 1000, and determines whether there is a deviation exceeding a predetermined value (step S10). If there is no deviation exceeding the predetermined value (step S10 → No), the process ends. On the other hand, if there is a deviation exceeding the predetermined value (step S10 → Yes), the process proceeds to the next step S11.

[0130] In step S11, the power cost efficiency optimal solution update unit 160 substitutes the room initial temperature (actual measurement value) of the latest corresponding turn into the power consumption cost efficiency calculation formula (Formula 3) and updates the optimal solution using a pruning method. In other words, the power cost efficiency optimal solution update unit 160 applies the same pruning method (pruning process) as that performed by the power cost efficiency optimal solution calculation unit 151 to the power efficiency calculations from the re-calculation turn to the Nth turn, thereby calculating the power consumption cost efficiency (E).

[0131] Here, if the method (method 1) of recalculating only for the room 500 in which a temperature deviation has occurred is set, the power cost efficiency optimal solution update unit 160 substitutes the latest actual measurement value into the formula for power consumption cost efficiency (E) only for the room 500 in which a temperature deviation has occurred (see formula (4) in Figure 16), and updates the air conditioning control level for the following turn.

[0132] Furthermore, if a method (method 2) of recalculating for all rooms 500 is set when a temperature deviation occurs in any of the rooms 500, the power cost efficiency optimal solution update unit 160 substitutes the latest actual measured value for all rooms 500 into the formula for power consumption cost efficiency (E) (see formula (5) in Figure 17) and updates the air conditioning control level for the following turn.

[0133] Then, in the turn following the turn in which the calculation of the optimal solution is completed, the power cost efficiency optimal solution update unit 160 executes control of the air conditioning control level of each room 500 in the allocation arrangement pattern of the newly calculated optimal solution (step S12).

[0134] By doing this, the power consumption control device 10 uses the pruning method to recalculate the power consumption cost efficiency (E) that is more closely matched to the actual measured value, thereby making it possible to further improve the power consumption cost efficiency of the entire DC1000 during the operational phase and reduce the calculation time for the optimal solution that maximizes the power consumption cost efficiency.

[0135] <Modification of the Present Embodiment> A modification of the power amount control device 10 according to the present embodiment will be described. In the power amount control device 10 according to the present embodiment, the load allocation control within the DC 1000 (load allocation patterns between rooms, load allocation patterns within the rooms) and the air conditioning control level of each room are taken into consideration, and then, based on information on fluctuations in the total load amount of the DC 1000 and fluctuations in electricity rates, an allocation allocation pattern and an air conditioning control level for each room 500 that maximize the power consumption cost efficiency (E) of the entire DC 1000 are determined as optimal solutions. In contrast, in the power amount control device 10 according to the modification, the load allocation amount to each room and the allocation allocation pattern within each room are fixed for each turn, and the load allocation control within the DC 1000, i.e., the allocation allocation pattern, is not taken into consideration.

[0136] In this case, the air conditioning power / temperature function approximation unit 130 approximates the air conditioning power function and the temperature function when the air conditioning control level is changed for each room 500 in a uniquely fixed distribution arrangement pattern. Then, the optimal control calculation unit 150 (power cost efficiency optimal solution calculation unit 151) uses a pruning method to calculate the power consumption cost efficiency (E) of the entire DC 1000 over N turns when the air conditioning control level is changed for each room 500 in the distribution arrangement pattern for each room that is fixed in each turn, and determines the air conditioning control level for each room 500 that maximizes the power consumption cost efficiency (E).

[0137] By doing this, the power consumption control device 10 according to the modified example of this embodiment can determine the air conditioning control level for each room that maximizes the power consumption cost efficiency even when there is no need to consider the load placement control within DC 1000, specifically the distribution placement pattern, and can reduce the calculation time for the optimal solution that maximizes the power consumption cost efficiency.

[0138] As described above, the power consumption control device 10 according to this embodiment can reduce the calculation time for the optimal solution that maximizes the power consumption cost efficiency by using a pruning technique when calculating the power consumption cost efficiency.

[0139] As a specific example of the calculation of the power consumption cost efficiency (E) by the power consumption control device 10 of this embodiment, the calculation processing time on a laptop PC for a scenario with 7 to 13 turns for one room, a fixed allocation pattern, and a total air-conditioning control level of "6" is shown. The optimal solution was calculated for the future 6 hours (12 turns), with one turn being 30 minutes, every 6 hours. As shown in Figure 20, when the power consumption cost efficiency (E) is calculated by the power consumption control device 10 using the pruning method, there is no dramatic increase in calculation processing time even with an increase in the number of turns. In this simulation, based on the number of processing patterns and the measured time when the number of turns was "7," the predicted calculation time was 0.10 seconds when the number of turns was "7," but was 0.17 seconds when the number of turns was "12" (6 hours). Therefore, a significant reduction in calculation time is possible compared to the calculation results without the pruning method, which showed a predicted calculation time of approximately 518 hours for the number of turns "12," as shown in Figure 22.

[0140] <Hardware Configuration> The power amount control device 10 and the room control device 20 according to this embodiment are realized by a computer 900 configured as shown in Fig. 21. Fig. 21 is a hardware configuration diagram showing an example of the computer 900 that realizes the functions of the power amount control device 10 and the room control device 20 according to this embodiment. The computer 900 has a CPU (Central Processing Unit) 901, a ROM (Read Only Memory) 902, a RAM 903, an HDD (Hard Disk Drive) 904, an input / output I / F (Interface) 905, a communication I / F 906, and a media I / F 907.

[0141] The CPU 901 operates based on programs stored in the ROM 902 or the HDD 904, and performs control by the control unit. The ROM 902 stores a boot program executed by the CPU 901 when the computer 900 is started up, programs related to the hardware of the computer 900, and the like.

[0142] The CPU 901 controls an input device 910 such as a mouse or keyboard, and an output device 911 such as a display or printer, via an input / output I / F 905. The CPU 901 acquires data from the input device 910 via the input / output I / F 905, and outputs generated data to the output device 911. Note that a GPU (Graphics Processing Unit) or the like may be used as a processor together with the CPU 901.

[0143] The HDD 904 stores programs executed by the CPU 901 and data used by the programs. The communication I / F 906 receives data from other devices via a communication network (e.g., NW (Network) 920) and outputs the data to the CPU 901, and also transmits data generated by the CPU 901 to other devices via the communication network.

[0144] The media I / F 907 reads a program or data stored in the recording medium 912 and outputs it to the CPU 901 via the RAM 903. The CPU 901 loads a program related to a target process from the recording medium 912 onto the RAM 903 via the media I / F 907, and executes the loaded program. The recording medium 912 is an optical recording medium such as a DVD (Digital Versatile Disc) or a PD (Phase Change Rewritable Disc), a magneto-optical recording medium such as an MO (Magneto Optical Disc), a magnetic recording medium, a semiconductor memory, or the like.

[0145] For example, when the computer 900 functions as the power amount control device 10 and the in-room control device 20 of the present invention, the CPU 901 of the computer 900 executes a program loaded onto the RAM 903 to realize the functions of the power amount control device 10 and the in-room control device 20. In addition, the HDD 904 stores data in the RAM 903. The CPU 901 reads and executes a program related to a target process from the recording medium 912. Alternatively, the CPU 901 may read a program related to a target process from another device via a communication network (NW 920).

[0146] <Effects> The effects of the power amount control device 10 and the like according to the present invention will be described below. The power amount control device according to the present invention is a power amount control device (10) that is communicatively connected to in-room control devices (20) for each room that control servers (3) and air conditioners (4) in rooms (500) in a data center (1000), and that allocates the amount of load to be processed in each room (500). In each room (500), a plurality of servers (3) are arranged in each placement control area (30) that indicates a consolidated area for processing the load, and one or more air conditioners (4) are arranged. The power amount control device (10) controls the servers (3) and air conditioners (4) in a turn that indicates a predetermined control time for a learning pattern that combines a plurality of stages of initial temperature in the room (500), a plurality of stages of server load to be loaded on the entire room, a plurality of load placement patterns that indicate the ratio at which the server load is distributed and allocated to the placement control areas (30), and a plurality of stages of air conditioning control levels of the air conditioners (4). The power amount control device (10) acquires the amount of air conditioning power consumption of the entire room and the temperature in the room at the end of the turn from the in-room control devices (20) for each room, as a learning history for each learning pattern, and generates learning history information (550) for each room. The air conditioning power consumption in the room relative to the server load allocated to each room is defined as an air conditioning power function, and in each room, for each load placement pattern and air conditioning control level, the air conditioning power function is approximated based on learning history information, and the temperature in the room at the end of the turn relative to the load allocated to the room is defined as a temperature function. In each room, for each load placement pattern and air conditioning control level, the air conditioning power / temperature function approximation unit approximates the temperature function based on the learning history information, and the total sum of the electricity charges based on the air conditioning power consumption in each successive turn and the electricity charges based on the server load in each successive turn is divided by the total load in each successive turn, and the power consumption cost efficiency of the entire data center is calculated when the air conditioning control level for each room is changed while the load allocation pattern for allocating the total load to the data center in each successive turn to each room and the load placement pattern in the room are fixed.and a power cost efficiency optimal solution calculation unit 151 that uses approximated air conditioning power functions and temperature functions to calculate, as an optimal solution, the air conditioning control level for each room 500 that maximizes the power consumption cost efficiency throughout the data center 1000, and the power cost efficiency optimal solution calculation unit 151 is characterized in that, when calculating the power consumption cost efficiency, at the end of each turn, it retains only one pattern of air conditioning control level with the lowest total electricity charge up to that turn for each temperature level that is a plurality of stages of room temperature that can be attained at the end of the turn, and uses that as the initial temperature level for the next turn and the total electricity charge solution up to that turn that corresponds to the temperature at the end of the turn.

[0147] In this way, the power amount control device 10 can calculate an optimal solution that maximizes the power consumption cost efficiency of the entire data center 1000 for the air conditioning control level for each turn with the load distribution pattern between each room 500 of the data center 1000 and the load allocation pattern within the room 500 fixed. Therefore, the power amount control device 10 can improve the power consumption cost efficiency of the entire DC 1000. Furthermore, when calculating the power consumption cost efficiency, the power amount control device 10 can significantly reduce the calculation time for the optimal solution by performing a pruning process that uses only the branches of the air conditioning control level with the lowest total electricity charge up to the corresponding turn for each temperature level in each turn.

[0148] The power amount control device according to the present invention is a power amount control device 10 that is communicatively connected to a room control device 20 for each room that controls a server 3 and an air conditioner 4 in a room 500 in a data center 1000, and that allocates the load amount to be processed in each room 500. In the room 500, a plurality of servers 3 are arranged in each placement control area 30 that indicates a grouped area for processing the load, and one or more air conditioners 4 are arranged, and a plurality of stages of initial temperature in the room, a plurality of stages of server load amount to be loaded on the entire room, and a plurality of stages of server load amount to be loaded on the placement control area 30 are set. The learning history information generating unit 120 controls the server 3 and the air conditioners 4 in a turn indicating a predetermined control time for a learning pattern, each of which is a combination of a plurality of load allocation patterns indicating the ratio at which loads are distributed and allocated, and a plurality of stages of air conditioning control levels for the air conditioners, and obtains the air conditioning power consumption of the entire room and the temperature in the room at the end of the turn from the in-room control device 20 for each room as a learning history for each learning pattern, and generates learning history information 550 for each room. The amount of power consumption is defined as an air conditioning power function, and in each room 500, the air conditioning power function is approximated based on learning history information 550 for each load placement pattern and air conditioning control level, and the room temperature at the end of the turn for the amount of load allocated to the room 500 is defined as a temperature function, and in each room 500, the air conditioning power / temperature function approximation unit 130 approximates the temperature function based on learning history information 550 for each load placement pattern and air conditioning control level, and calculates the total load on the data center 1000 for each successive turn for each room 500. A load allocation pattern generation unit 140 generates an allocation pattern by changing and combining a load allocation pattern to be allocated and the load allocation pattern in a room, and defines the power consumption cost efficiency as the value obtained by dividing the total sum of the electricity charges based on the air conditioning power consumption amount for each successive turn and the total electricity charges based on the server load amount for each successive turn by the total load amount for each successive turn, and calculates the power consumption cost efficiency of the entire data center when the air conditioning control level for each room is changed for each generated allocation pattern as follows:The data center includes an optimal control calculation unit 150 that uses approximated air conditioning power functions and temperature functions to calculate an allocation arrangement pattern that maximizes the power consumption cost efficiency throughout the data center and an optimal solution for the air conditioning control level of each room 500, and the optimal control calculation unit 150 is characterized in that, when calculating the power consumption cost efficiency, at the end of each turn, the optimal control calculation unit 150 retains only one pattern of air conditioning control level that has the lowest total electricity charge up to that turn for each temperature level that is a plurality of stages of room temperature that can be attained at the end of the turn, and performs pruning processing that indicates that the initial temperature level of the next turn corresponding to the temperature at the end of the turn and the total electricity charge solution up to that turn are to be retained.

[0149] In this way, the power amount control device 10 can calculate an optimal solution that maximizes the power consumption cost efficiency of the entire data center 1000 for the air conditioning control level in the load distribution pattern between each room 500 of the data center 1000 and the load allocation pattern within the room 500. Therefore, the power amount control device 10 can improve the power consumption cost efficiency of the entire DC 1000. Furthermore, when calculating the power consumption cost efficiency, the power amount control device 10 can significantly reduce the calculation time for the optimal solution by performing a pruning process in each turn that uses only the branch of the air conditioning control level that has the lowest total electricity charge up to the corresponding turn for each temperature level.

[0150] Furthermore, the power amount control device 10 is characterized by further comprising a power cost efficiency optimal solution update unit 160 that, during operation with the calculated optimal solution, acquires initial temperature information from each room control device 20 as an actual measurement value at the beginning of each turn, compares the actual measurement value with the initial temperature of the turn indicated by the room temperature at the end of the turn immediately preceding the turn, which is calculated as the optimal solution, and, if there is a deviation exceeding a predetermined value, recalculates the power consumption cost efficiency from the turn in which the deviation was found using the actual measurement value through pruning processing, thereby updating the optimal solution.

[0151] In this way, the power amount control device 10 acquires an actual measurement value, which is the initial temperature in each room, at the beginning of each turn during operation, compares the actual measurement value with the initial temperature for that turn calculated as the optimal solution, and if there is a discrepancy exceeding a predetermined value, substitutes the actual measurement value into the calculation formula for power consumption cost efficiency to recalculate the optimal solution. The power amount control device 10 can further improve power consumption cost efficiency and improve the effect of reducing electricity charges. Furthermore, by using a pruning process during this recalculation, the power amount control device 10 can significantly reduce the calculation time for the optimal solution.

[0152] The present invention is not limited to the above-described embodiments, and many modifications can be made by a person having ordinary skill in the art within the technical concept of the present invention.

[0153] REFERENCE SIGNS LIST 1 Power amount control system 3 Server 4 Air conditioner 10 Power amount control device 20 In-room control device 30 Placement control area 40 Air conditioning control area 100, 200 Control unit 110 Room definition unit 120 Per-room learning history information generation unit 130 Air conditioning power / temperature function approximation unit 140 Allocation placement pattern generation unit 150 Optimal control calculation unit 151 Power cost efficiency optimal solution calculation unit 152 Control pattern determination unit 160 Power cost efficiency optimal solution update unit 170, 250 Input / output unit 180, 260 Memory unit 210 Temperature measurement unit 220 Learning history collection unit 230 In-room server control unit 231 Load placement pattern setting unit 232 Server power measurement unit 240 In-room air conditioning control unit 241 Air conditioning control execution unit 242 Air conditioning power measurement unit 500 Room 550 Learning history information 1000 Data center (DC)

Claims

1. An electric power control device that is communicatively connected to an in-room control device for each room that controls servers and air conditioners in the rooms in a data center and allocates the load to be processed in each of the rooms, wherein the rooms have a plurality of servers and one or more air conditioners arranged in each placement control area that indicates a consolidated area for processing the load, the device comprising: a per-room learning history information generation unit that controls the servers and the air conditioners in a turn indicating a predetermined control time for learning patterns that combine a plurality of stages of initial temperatures in the room, a plurality of stages of server loads to be loaded on the entire room, a plurality of load placement patterns that indicate the ratio at which the server loads are distributed and placed in the placement control areas, and a plurality of stages of air conditioning control levels of the air conditioners, and obtains the air conditioning power consumption of the entire room and the in-room temperature at the end of the turn from the in-room control device for each room as a learning history for each learning pattern, and generates learning history information for each room; an air conditioning power / temperature function approximation unit that defines the amount of air conditioning power consumption in the room relative to the amount of server load allocated to the room as an air conditioning power function, and in each of the rooms, approximates the air conditioning power function based on the learning history information for each of the load placement patterns and the air conditioning control levels, and defines the temperature in the room at the end of the turn relative to the amount of load allocated to the room as a temperature function, and in each of the rooms, approximates the temperature function based on the learning history information for each of the load placement patterns and the air conditioning control levels;a power cost efficiency optimal solution calculation unit that defines a power consumption cost efficiency as a value obtained by dividing the total value of a total of an electricity charge based on the air conditioning power consumption of each successive turn and a total of an electricity charge based on the server load of each successive turn by the total load of each successive turn, calculates the power consumption cost efficiency of the entire data center when the air conditioning control level for each room is changed in a state in which a load allocation pattern for allocating a total load amount to the data center for each successive turn to each room and the load allocation pattern within the room are fixed, using the approximated air conditioning power function and the temperature function, and calculates the air conditioning control level of each room that maximizes the power consumption cost efficiency of the entire data center as an optimal solution, wherein the power cost efficiency optimal solution calculation unit performs a pruning process indicating that, when calculating the power consumption cost efficiency, at the end of each turn, only one pattern of air conditioning control level with the lowest total electricity charge up to that turn is retained for each temperature level that is a plurality of stages of room temperature that can be attained at the end of the turn, and sets the initial temperature level of the next turn corresponding to the temperature at the end of the turn and the total electricity charge solution up to that turn. An electric power control device comprising:

2. A power consumption control device that is communicatively connected to an in-room control device for each room that controls servers and air conditioners in the rooms in a data center and allocates the load to be processed in each of the rooms, wherein the rooms have a plurality of servers and one or more air conditioners arranged in each placement control area that indicates a consolidated area for processing the load, and a per-room learning history information generation unit that controls the servers and the air conditioners in a turn indicating a predetermined control time for learning patterns that combine a plurality of stages of initial temperatures in the room, a plurality of stages of server loads to be loaded on the entire room, a plurality of load placement patterns that indicate the ratio at which the server loads are distributed and placed in the placement control areas, and a plurality of stages of air conditioning control levels of the air conditioners, thereby acquiring the air conditioning power consumption of the entire room and the in-room temperature at the end of the turn from the in-room control device for each room, as a learning history for each learning pattern, and generates learning history information for each room; an air conditioning power / temperature function approximation unit that defines the amount of air conditioning power consumption in the room relative to the amount of server load allocated to the room as an air conditioning power function, and in each of the rooms, for each of the load placement patterns and the air conditioning control levels, approximates the air conditioning power function based on the learning history information, and defines the temperature in the room at the end of the turn relative to the amount of load allocated to the room as a temperature function, and in each of the rooms, approximates the temperature function based on the learning history information for each of the load placement patterns and the air conditioning control levels; an allocation placement pattern generation unit that generates an allocation placement pattern that changes and combines a load allocation pattern that allocates a total load amount to the data center for each of successive turns to each room and the load allocation pattern in the room;an optimal control calculation unit that defines a power consumption cost efficiency as the sum of a total electricity charge based on the air conditioning power consumption of each successive turn and a total electricity charge based on the server load of each successive turn divided by the total load of each successive turn, calculates the power consumption cost efficiency of the entire data center when the air conditioning control level for each room is changed for each generated distribution arrangement pattern using the approximated air conditioning power function and the temperature function, and calculates the distribution arrangement pattern and the air conditioning control level of each room that maximize the power consumption cost efficiency of the entire data center as an optimal solution, wherein the optimal control calculation unit performs a pruning process that indicates, when calculating the power consumption cost efficiency, that at the end of each turn, only one pattern of air conditioning control level with the lowest total electricity charge up to that turn for each temperature level that is a plurality of stages of room temperature that can be obtained at the end of the turn, and sets the initial temperature level of the next turn corresponding to the temperature at the end of the turn and the total electricity charge solution up to that turn.

3. The power amount control device according to claim 1 or 2, further comprising a power cost efficiency optimal solution update unit that, during operation with the calculated optimal solution, obtains initial temperature information from each in-room control device as an actual measurement value at the beginning of each turn, compares the actual measurement value with the initial temperature of the turn indicated by the in-room temperature at the end of the turn immediately preceding the turn calculated as the optimal solution, and, if there is a deviation exceeding a predetermined value, recalculates the power consumption cost efficiency using the actual measurement value from the turn in which the deviation was found using the pruning process, and updates the optimal solution.

4. An electric power control program for causing a computer to function as the electric power control device according to claim 1 or 2.

Images