Facility control device, facility control method, and facility control program

Abstract

A facility control device (100) for supporting decision-making regarding facility control settings for a target building is provided with a data acquisition unit (190), an optimization calculation unit (110), and an extraction unit (140). The data acquisition unit (190) acquires other-building data, which is preference-based facility control setting data for other buildings different from the target building. The optimization calculation unit (110) uses an optimization technique to create a plurality of facility control setting candidates within a facility controllable range for the target building as elements of a created candidate group. The extraction unit (140) extracts, from the created candidate group, one or more facility control setting candidates as elements of an extracted candidate group, using an inference model (131), which is a model based on the other-building data and infers facility control setting preferences.

Description

Facility control device, facility control method, and facility control program

[0001] The present disclosure relates to an equipment control device, an equipment control method, and an equipment control program.

[0002] A system is known that extracts multiple candidate equipment control settings using an optimization technique such as mathematical programming and supports decision-making regarding equipment control settings based on the extracted multiple candidate equipment control settings. Patent Document 1 discloses a method for visualizing conflicts between constraints as a method for selecting one equipment control setting from multiple equipment control settings. Patent Document 2 discloses a method for extracting some candidate equipment control settings based on conditions such as proximity to a certain value or a certain interval, and presenting them to a decision maker.

[0003] JP 2021-144569 A Patent No. 6073000 A

[0004] The method disclosed in Patent Document 1 has a problem in that the decision maker's implicit decision-making criteria must be considered as constraints before optimization. The method disclosed in Patent Document 2 has a problem in that the decision maker's preferences cannot be sufficiently reflected, and there is a risk that equipment control settings that deviate from the decision maker's preferences will continue to be presented to the decision maker. Therefore, according to the prior art, when the decision maker's preferences are unknown, it is not possible to reflect the decision maker's preferences and make a decision that the decision maker can accept with little effort.

[0005] The present disclosure aims to extract equipment control settings relatively efficiently in support of decision-making regarding equipment control settings even when the preferences of a decision maker are unknown.

[0006] The equipment control device according to the present disclosure is an equipment control device that supports decision-making regarding equipment control settings for a target building, and includes: a data acquisition unit that acquires other building data, which is data based on equipment control setting preferences for other buildings different from the target building; an optimization calculation unit that uses optimization technology to create multiple equipment control setting candidates as elements of a group of candidates to be created within the equipment controllable range for the target building; and an extraction unit that uses an inference model, which is a model based on the other building data and is a model that infers equipment control setting preferences, to extract one or more equipment control setting candidates from the group of candidates to be created as elements of a group of candidates to be extracted.

[0007] According to the present disclosure, an extractor extracts one or more candidate equipment control settings from a plurality of candidate equipment control settings created using optimization techniques using an inference model that is a model based on other building data and that infers preferences for equipment control settings. Therefore, according to the present disclosure, in support of decision-making regarding equipment control settings, it is possible to extract equipment control settings relatively efficiently even when the preferences of a decision maker are unknown.

[0008] FIG. 1 is a diagram showing an example of the configuration of an equipment control system 90 according to a first embodiment. FIG. 2 is a diagram showing an example of the hardware configuration of an equipment control device 100 according to the first embodiment. FIG. 3 is a flowchart showing the operation of the equipment control device 100 according to the first embodiment during inference. FIG. 4 is a diagram explaining the processing of the optimization calculation unit 110 and the extraction unit 140 according to the first embodiment, where (a) is a diagram showing equipment control setting candidates created by the optimization calculation unit 110, and (b) is a diagram showing equipment control setting candidates extracted by the extraction unit 140. FIG. 5 is a flowchart showing the operation of the equipment control device 100 according to the first embodiment during learning. FIG. 6 is a diagram showing an example of the hardware configuration of an equipment control device 100 according to a modified example of the first embodiment. FIG. 7 is a diagram showing an example of the configuration of an equipment control system 90 according to a second embodiment. FIG. 8 is a diagram showing an example of the configuration of an equipment control system 90 according to a third embodiment. FIG. 9 is a diagram showing an example of the configuration of an equipment control system 90 according to a modified example of the third embodiment. FIG. 10 is a diagram showing an example of the configuration of an equipment control system 90 according to a modified example of the third embodiment. FIG. 11 is a diagram showing an example of the configuration of an equipment control system 90 according to a modified example of the third embodiment. 10A and 10B are diagrams illustrating the processing of the optimization calculation unit 110 and the extraction unit 140 according to the fourth embodiment, in which FIG. 10A shows candidate equipment control settings created by the optimization calculation unit 110, and FIG. 10B shows candidate equipment control settings extracted by the extraction unit 140.

[0009] In the description of the embodiments and drawings, the same elements and corresponding elements are given the same symbols. The description of elements given the same symbols is omitted or simplified as appropriate. Arrows in the drawings mainly indicate the flow of data or the flow of processing. Furthermore, "unit" may be interpreted as "circuit," "device," "equipment," "process," "step," "procedure," "processing," or "circuitry" as appropriate. The functions of each unit provided in each device may be realized by firmware, software, hardware, or a combination of these.

[0010] First Embodiment Hereinafter, the present embodiment will be described in detail with reference to the drawings.

[0011] ***Configuration*** FIG. 1 shows an example of the configuration of an equipment control system 90 according to this embodiment. The equipment control system 90 includes an equipment control device 100, an equipment management device 20, a display device 30, and an input device 40. The multiple devices included in the equipment control system 90 may be integrated as appropriate. The equipment control device 100 is a device that supports decision-making regarding equipment control settings for a target building and determines the control settings for each piece of equipment according to the preferences of an equipment manager. As shown in FIG. 1 , the equipment control device 100 includes an optimization calculation unit 110, a storage unit 120, a learning unit 130, an extraction unit 140, a display unit 150, an input unit 160, a selection unit 170, a communication unit 180, and a data acquisition unit 190. The equipment control device 100 utilizes other building data 191. The equipment management device 20 is a device that manages equipment in accordance with instructions from the equipment control device 100. The display device 30 is a device that displays output from the equipment control device 100 to the equipment manager. The input device 40 has a function of receiving input from an equipment manager and transmitting the received input to the equipment control device 100 .

[0012] The optimization calculation unit 110 uses optimization techniques to calculate multiple candidate equipment control settings that meet the preferences of the equipment manager within the equipment controllable range for the target building. The multiple candidate equipment control settings may be considered as elements of a candidate set. In this case, the optimization calculation unit 110 may use data acquired for each piece of equipment and its surroundings, as well as other data related to the operation of each piece of equipment, such as data indicating date and time. The equipment control settings may indicate an operation plan for each piece of equipment.

[0013] The storage unit 120 stores each past selection result by the facility manager and each condition corresponding to each selection result. The storage unit 120 may record, as learning data, data linking a group of extraction candidates with elements selected by the facility manager from the group of extraction candidates.

[0014] The learning unit 130 learns the inference model 131 based on each past selection result by the facility manager and each condition corresponding to each selection result. The learning unit 130 may re-learn the inference model 131 based on the learning data.

[0015] The extraction unit 140 extracts equipment control setting candidates that suit the preferences of the equipment manager from the equipment control setting candidates calculated by the optimization calculation unit 110, through inference using the inference model 131. That is, the extraction unit 140 uses the inference model 131 to extract one or more equipment control setting candidates from the group of candidates to be created as elements of the group of candidates to be extracted.

[0016] The display unit 150 displays the equipment control setting candidates extracted by the extraction unit 140, i.e., each element of the extracted candidate group, to the equipment manager of the target building.

[0017] The input unit 160 accepts input of equipment control settings selected by the equipment manager.

[0018] The selection unit 170 selects equipment control settings to be actually applied to each piece of equipment from the equipment control setting candidates extracted by the extraction unit 140 based on the input received by the input unit 160 .

[0019] The communication unit 180 transmits the equipment control settings selected by the selection unit 170 to the equipment management device 20 .

[0020] The data acquisition unit 190 acquires other building data 191 from outside the device.

[0021] The other building data 191 is data based on preferences for equipment control settings for other buildings different from the target building. The other building data 191 may be data for learning the inference model 131. Specific examples of the other building data 191 include data showing the selection results of an equipment manager for other buildings, or an existing inference model learned using the data. The other building is one or more buildings other than the target building. The target building is a building equipped with each piece of equipment that is the target of control by the equipment control device 100.

[0022] The inference model 131 will be described in detail below. The inference model 131 may be a model based on the other building data 191 and may infer preferences for equipment control settings. First, the result of a single selection of an equipment control setting by an equipment manager includes, as specific examples, data indicating energy consumption, satisfaction, and the equipment control setting as each piece of feature data. Each piece of feature data is treated as a feature vector. As specific examples of energy consumption, the data indicates at least one of annual power consumption [kWh], a Building Energy Index (BEI) value, and a percentage range of the controllable range of energy consumption. As specific examples of satisfaction, the data indicates at least one of several satisfaction-related indices and a percentage range of the controllable range of satisfaction. As the satisfaction index, various proposed indices may be used, such as an index related to thermal sensation, such as PMV (Predicted Mean Vote), or the environmental satisfaction index presented in [Reference 1]. Specific examples of the equipment control settings include the dimming rate [%] of lighting for each room, the set temperature [°C] of the air conditioner for each room, and the difference [°C] between the set temperatures for each room.

[0023] [Reference 1] K. Fukuhara et al. , “Digital Twin Based Evolutionary Building Facility Control Optimization,”2022 IEEE Congress on Evolutionary Computation (CEC), 2022, pp. 1-8

[0024] The learning unit 130 trains the inference model 131 by modeling the preferences (profile) of the equipment manager using feature data based on multiple selection results by the equipment manager. The inference model 131 may utilize a recommendation method, such as cosine similarity or a naive Bayes classifier, employed in content-based filtering of recommendations on electronic commerce (EC) sites, as a specific example, or other methods. When the inference model 131 utilizes cosine similarity, the closeness of each equipment control setting candidate to the equipment manager's profile is obtained. Therefore, in this case, a predetermined number of candidates may be extracted and output from the equipment control setting candidates in order of corresponding closeness. The similarity may be calculated after mapping multidimensional features to two dimensions using a self-organization map (SOM) or a variational autoencoder (VAE). When a classifier is used in the inference model 131, it is determined whether each candidate equipment control setting matches the equipment manager profile. Therefore, in this case, only equipment control settings that match the equipment manager profile may be extracted and output. Furthermore, if the results of equipment control setting selections by equipment managers in other buildings can also be used, methods used in collaborative filtering, in addition to content-based methods, may be used. Furthermore, when selecting an equipment control setting, it is easier to select from candidates with significantly different energy consumption, satisfaction, etc. Therefore, the candidate equipment control settings may be divided into several clusters in advance, an inference model 131 may be constructed for each cluster, and a set of equipment control settings recommended by the inference model 131 corresponding to each cluster may be output as extracted candidates. As a specific example, with regard to the clusters, ranges corresponding to each cluster may be defined in advance by dividing the range of energy consumption in increments of N units, and each cluster may be constructed consisting of candidate equipment control settings corresponding to each range.The target of range segmentation may be satisfaction or equipment control settings instead of energy consumption, or a combination of these. Clusters may also be constructed by grouping data according to the proximity of corresponding features using a k-means method or the like. Features may also be mapped to two dimensions using a SOM or VAE, and then range segmentation or k-means classification may be performed.

[0025] 2 shows an example of the hardware configuration of the equipment control device 100 according to this embodiment. The equipment control device 100 is made up of a computer. The equipment control device 100 may be made up of multiple computers.

[0026] As shown in the figure, the equipment control device 100 is a computer that includes hardware such as a processor 11, a memory 12, an auxiliary storage device 13, an input / output interface (IF) 14, and a communication device 15. These pieces of hardware are connected as appropriate via signal lines 19.

[0027] The processor 11 is an integrated circuit (IC) that performs arithmetic processing and controls the hardware of a computer. Specific examples of the processor 11 include a central processing unit (CPU), a digital signal processor (DSP), or a graphics processing unit (GPU). The equipment control device 100 may include multiple processors that replace the processor 11. The multiple processors share the role of the processor 11.

[0028] The memory 12 is typically a volatile storage device, specifically a random access memory (RAM). The memory 12 is also called a primary storage device or a main memory. Data stored in the memory 12 is saved in the secondary storage device 13 as needed.

[0029] The auxiliary storage device 13 is typically a non-volatile storage device, and specific examples thereof include a ROM (Read Only Memory), an HDD (Hard Disk Drive), or a flash memory. Data stored in the auxiliary storage device 13 is loaded into the memory 12 as needed. The memory 12 and the auxiliary storage device 13 may be configured integrally.

[0030] The input / output IF 14 is a port to which an input device and an output device are connected. Specific examples of the input / output IF 14 include a USB (Universal Serial Bus) terminal. Specific examples of the input device include a keyboard and a mouse. Specific examples of the output device include a display.

[0031] The communication device 15 is a receiver and a transmitter, and is specifically a communication chip or a NIC (Network Interface Card).

[0032] Each unit of the equipment control device 100 may use the input / output IF 14 and the communication device 15 as appropriate when communicating with other devices.

[0033] The auxiliary storage device 13 stores an equipment control program. The equipment control program is a program that causes a computer to realize the functions of each unit included in the equipment control device 100. The equipment control program is loaded into the memory 12 and executed by the processor 11.

[0034] Data used when executing the equipment control program and data obtained by executing the equipment control program are stored in a storage device as appropriate. Each part of the equipment control device 100 uses a storage device as appropriate. Specific examples of the storage device include at least one of the memory 12, the auxiliary storage device 13, a register in the processor 11, and a cache memory in the processor 11. Note that the terms "data" and "information" may have the same meaning. The storage device may be independent of the computer. The functions of the memory 12 and the auxiliary storage device 13 may be realized by other storage devices.

[0035] The equipment control program may be recorded on a computer-readable non-volatile recording medium. Specific examples of the non-volatile recording medium include an optical disk and a flash memory. The equipment control program may be provided as a program product.

[0036] ***Description of Operation*** The operation procedure of the equipment control device 100 corresponds to an equipment control method. Also, the program that realizes the operation of the equipment control device 100 corresponds to an equipment control program.

[0037] 3 is a flowchart showing an example of the operation during inference of the equipment control device 100. This operation will be described with reference to FIG.

[0038] (Step S101) The optimization calculation unit 110 creates a plurality of candidates for equipment control settings.

[0039] (Step S102) The extraction unit 140 uses the inference model 131 to extract equipment control settings that match the preferences of the equipment manager from among multiple equipment control setting candidates. Here, the inference model 131 is a model that learns the preferences of the equipment manager. Note that when a certain equipment manager uses the equipment control device 100 for the first time, the preferences of the certain equipment manager are unknown. Therefore, the equipment control device 100 acquires other building data 191 and prepares the inference model 131 using the acquired other building data 191. Furthermore, by executing the learning operation described below, after the equipment manager's selection result for the target building is obtained, the inference model 131 is trained using data indicating the selection result, and inference and extraction are performed based on the trained inference model 131.

[0040] Fig. 4 is a diagram illustrating the processing of the optimization calculation unit 110 and the extraction unit 140. Fig. 4(a) shows an example of a plurality of equipment control setting candidates created by the optimization calculation unit 110. Fig. 4(a) shows each equipment control setting candidate created by the optimization calculation unit 110 within the equipment controllable range, taking satisfaction level and energy consumption into consideration. Fig. 4(b) is a diagram corresponding to Fig. 4(a) and shows an example of equipment control setting candidates extracted by the extraction unit 140.

[0041] (Step S103) The display unit 150 displays only the equipment control settings extracted by the extraction unit 140 to the equipment manager.

[0042] (Step S104) The equipment manager selects one equipment control setting from the equipment control setting candidates displayed on the display device 30, and inputs the selected equipment control setting from the input device 40. The input unit 160 accepts the input from the input device 40. When an equipment control setting is selected, the equipment control device 100 transitions to step S105.

[0043] (Step S105) The selection unit 170 selects one equipment control setting from the candidate equipment control settings based on the input received by the input unit 160, and transmits the selected equipment control setting to the equipment management device 20 via the communication unit 180.

[0044] (Step S106) The selection unit 170 records, in the storage unit 120, a list of candidate equipment control settings and the equipment control setting selected from the list of candidate equipment control settings as the current selection result.

[0045] 5 is a flowchart showing an example of the operation during learning of the equipment control device 100. This operation will be described with reference to FIG.

[0046] (Step S111) If the selection result is recorded in the storage unit 120, the equipment control device 100 proceeds to step S112. Otherwise, the equipment control device 100 proceeds to step S115.

[0047] (Step S112) If the selection results recorded in the memory unit 120 satisfy a certain condition, the learning unit 130 proceeds to step S113 and starts learning using the selection results. Otherwise, the equipment control device 100 ends the processing of this flowchart. Here, the "certain condition" is, for example, one of the following conditions or a combination of two conditions: Condition 1: The number of selection results recorded in the memory unit 120 is equal to or greater than a certain number; Condition 2: An equipment control setting other than the equipment control setting extracted by the extraction unit 140 has been selected by the equipment manager.

[0048] (Step S113) The learning unit 130 uses the selection results recorded in the storage unit 120 to learn an inference model 131 that infers the preferences of the facility manager.

[0049] (Step S114) The learning unit 130 deploys the inference model 131 prepared based on other building data 191, or the inference model 131 learned using the selection results recorded in the memory unit 120, in a state that can be used by the extraction unit 140.

[0050] (Step S115) Before the selection result is recorded in the memory unit 120, the learning unit 130 prepares an inference model 131 based on the other building data 191. As a specific example, the learning unit 130 uses a means such as learning the inference model 131 based on the other building data 191, or using an inference model used in another building as the inference model 131 of the target building.

[0051] ***Explanation of Effects of First Embodiment*** When performing optimization calculations that explicitly consider the preferences implicitly held by an equipment manager, it is necessary to explicitly consider the preferences, for example, by formulating the preferences and incorporating them into the optimization calculations as objective functions and constraints. On the other hand, in this embodiment, the preferences of the equipment manager are learned based on the selection results of the equipment manager. Therefore, according to this embodiment, it is possible to select equipment control settings that reflect the preferences without explicitly considering the preferences.

[0052] Furthermore, when selecting equipment control settings, methods that extract settings based on predetermined conditions may result in the extraction conditions not matching the preferences of the equipment manager, making it impossible to present a reasonable solution and requiring the extraction to be started over again. In contrast, in this embodiment, the preferences of the equipment manager are learned based on the selection results of the equipment manager, and equipment control settings are extracted based on the learning results. Therefore, according to this embodiment, it is possible to extract and select settings that match the preferences of the equipment manager relatively efficiently. Furthermore, by utilizing this embodiment, the equipment manager can make satisfactory decisions with relatively little effort.

[0053] Furthermore, the preferences of an equipment manager are usually unknown before the inference model 131 is trained. In such a situation, it is difficult to extract equipment control settings that are at least somewhat suitable to the preferences of the equipment manager. On the other hand, in this embodiment, preferences are inferred using the inference model 131 trained using data including the selection results of other managers. Therefore, according to this embodiment, by appropriately selecting the selection results of other managers, it is possible to extract equipment control settings based on conditions that are relatively close to the preferences of the equipment manager, allowing the equipment manager to select equipment control settings relatively efficiently.

[0054] ***Other Configurations*** <Modification 1> Fig. 6 shows an example of the hardware configuration of the equipment control device 100 according to this modification. The equipment control device 100 includes a processing circuit 18 instead of the processor 11, the processor 11 and memory 12, the processor 11 and auxiliary storage device 13, or the processor 11, memory 12, and auxiliary storage device 13. The processing circuit 18 is hardware that realizes at least a portion of the components included in the equipment control device 100. The processing circuit 18 may be dedicated hardware, or may be a processor that executes a program stored in the memory 12.

[0055] When the processing circuitry 18 is dedicated hardware, the processing circuitry 18 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. The equipment control device 100 may be provided with multiple processing circuits that replace the processing circuitry 18. The multiple processing circuits share the role of the processing circuitry 18.

[0056] In the equipment control device 100, some functions may be realized by dedicated hardware, and the remaining functions may be realized by software or firmware.

[0057] The processing circuitry 18 is realized by, for example, hardware, software, firmware, or a combination of these. The processor 11, memory 12, auxiliary storage device 13, and processing circuitry 18 are collectively referred to as the "processing circuitry." In other words, the functions of the functional components of the equipment control device 100 are realized by the processing circuitry. Equipment control devices 100 according to other embodiments may also have a configuration similar to this modification.

[0058] Second Embodiment The following mainly describes the differences from the above-described embodiment with reference to the drawings.

[0059] 7 shows an example of the configuration of an equipment control system 90 according to this embodiment. The equipment control device 100 according to this embodiment further includes a data selection unit 210.

[0060] The other building data 191 according to this embodiment is data for training the inference model 131 and is data indicating each feature of the other building. Specific examples of each feature of the other building include the feature of the other building itself, or the feature of the facility manager of the other building. Specific examples of each feature include the feature of the location of the other building, the feature of the size of the other building (large / medium / small, etc.), the feature of the use of the other building (office / commercial facility / educational facility, etc.), the feature of the type of equipment the other building has (individually distributed air conditioning / central heat source, etc.), the feature of the age group of the facility manager of the other building, or the feature of the preference of the facility manager of the other building (emphasis on comfort / emphasis on energy conservation, etc.). The facility manager of the other building is the facility manager for the other building.

[0061] The data selection unit 210 selects data that matches each feature related to the target building from the data acquired by the data acquisition unit 190. That is, the data selection unit 210 refers to each feature related to the other buildings and selects data from the other building data 191 as selected data according to each feature related to the target building. Specific examples of each feature related to the target building include features related to the target building itself or features of the target facility manager. Each feature related to the target building is the same as each feature related to other buildings. Here, the target facility manager is the facility manager related to the target building. Facility manager can also be a general term for both the other facility managers and the target facility manager.

[0062] The learning unit 130 according to this embodiment learns the inference model 131 based on the data selected by the data selection unit 210, i.e., the selected data.

[0063] ***Explanation of Operation*** The following describes the differences between the operation of the equipment control device 100 according to the present embodiment during learning and the operation of the equipment control device 100 according to the first embodiment during learning.

[0064] (Step S115) This step is basically the same as step S115 in embodiment 1. However, the learning unit 130 learns the inference model 131 based on the data selected by the data selection unit 210.

[0065] ***Explanation of Effects of Embodiment 2*** In this embodiment, during learning, existing data is not randomly used, but data related to buildings similar to the target building, or data corresponding to preferences similar to those of the target equipment manager, etc. Therefore, according to this embodiment, it is possible to extract equipment control settings under conditions that are closer to the preferences of the target equipment manager, enabling the equipment manager to select equipment control settings more efficiently.

[0066] Third Embodiment Hereinafter, differences from the above-described embodiments will be mainly described with reference to the drawings.

[0067] 8 shows an example of the configuration of an equipment control system 90 according to this embodiment. The equipment control device 100 according to this embodiment further includes a data mixer 310.

[0068] The data acquisition unit 190 according to this embodiment acquires other building data 391 .

[0069] The other building data 391 is data for training the inference model 131, and is data indicating the selection results by other facility managers regarding other buildings.

[0070] The data mixing unit 310 appropriately mixes the other building data 391 with data indicating the selection result by the facility manager. Specifically, the data mixing unit 310 generates mixed data by mixing at least a part of the learning data with at least a part of the other building data 391.

[0071] The learning unit 130 according to this embodiment learns the inference model 131 based on the mixed data. That is, the learning unit 130 learns the inference model 131 using the data mixed by the data mixing unit 310, rather than using only either the other building data 391 or the data indicating the selection result by the facility manager.

[0072] ***Explanation of Operation*** The following describes the differences between the operation of the equipment control device 100 according to the present embodiment during learning and the operation of the equipment control device 100 according to the first embodiment during learning.

[0073] (Step S113) The learning unit 130 learns the inference model 131 using the data mixed by the data mixing unit 310 instead of the selection results recorded in the storage unit 120.

[0074] ***Explanation of the Effects of Embodiment 3*** In this embodiment, a mixture of data on other buildings or data on the target building is used when learning the inference model 131. Therefore, according to this embodiment, it is possible to construct an inference model 131 with relatively high accuracy even when there is little data on the target building.

[0075] 9 shows an example of the configuration of an equipment control system 90 according to this modification. The equipment control device 100 according to this modification further includes a data extraction unit 320.

[0076] The other building data 391 according to this modification is data indicating each feature of the other building.

[0077] The data extraction unit 320 extracts only data that matches each feature related to the target building from the other building data 391 acquired by the data acquisition unit 190. That is, the data extraction unit 320 extracts data that matches each feature related to the target building from the other building data 391 as extracted building data. Each feature is the same as each feature described in embodiment 2. In this modification, the inference model 131 is trained using the data extracted by the data extraction unit 320, rather than all data collected about other buildings.

[0078] The data mixer 310 according to this modification appropriately mixes the data extracted by the data extractor 320 with data indicating the selection result by the facility manager. That is, the data mixer 310 uses the extracted building data instead of the other building data 391.

[0079] ***Explanation of the Effects of Modification 2*** According to this modification, inference is performed using an inference model 131 that has been trained using data that matches the characteristics of the target building or the characteristics of the facility manager related to the target building. Therefore, according to this modification, even when there is little data related to the target building, the accuracy of inference can be improved.

[0080] <Modification 3> *** Description of Configuration *** Fig. 10 shows an example configuration of an equipment control system 90 according to this modification. This modification corresponds to an embodiment that expands on Modification 2. The equipment control device 100 according to this modification further includes a data expansion unit 330.

[0081] The data expansion unit 330 generates new data by combining data with different corresponding characteristics based on the data extracted by the data extraction unit 320. That is, the data expansion unit 330 generates data for training the inference model 131 based on the other building data 391 as data for expansion learning. The data expansion unit 330 generates data particularly when there is little data that matches each characteristic related to the equipment control device 100. As a specific example, the data expansion unit 330 generates new data by combining data that is similar only in the corresponding location area and use and data that is similar only in the corresponding scale and equipment type. In this case, the equipment control setting value selected by the equipment manager may be an intermediate value between these two data.

[0082] The data mixing unit 310 according to this modification appropriately mixes the data extracted by the data extraction unit 320, data indicating the selection results of the facility manager, and data generated by the data expansion unit 330. That is, the data mixing unit 310 generates mixed data by mixing at least a portion of the learning data, at least a portion of the other building data 391, and at least a portion of the extended learning data. In this modification, the data generated by the data expansion unit 330 is also used when training the inference model 131.

[0083] ***Explanation of the Effects of Modification Example 3*** In this modification example, even if there is little data that matches the characteristics of the target building or facility manager, the data is expanded based on the extracted other building data 391, and inference is performed based on the inference model 131 that has been trained based on the expanded data. Therefore, according to this modification example, it is possible to obtain inference results that are estimated to be somewhat close to the facility manager's preferences from an early stage.

[0084] Fourth Embodiment Hereinafter, differences from the above-described embodiments will be mainly described with reference to the drawings.

[0085] ***Description of Configuration*** Fig. 11 shows an example of the configuration of an equipment control system 90 according to this embodiment. The equipment control device 100 according to this embodiment further stores equipment characteristic data 431.

[0086] The equipment characteristic data 431, also called equipment operation characteristic data, is made up of data indicating the operation characteristics of each piece of equipment. Specific examples of the operation characteristics of each piece of equipment include the number of heat sources in operation, the tendency of air conditioning temperature settings, the tendency of fan air volume settings, and the tendency of operation times. The operation characteristics of equipment correspond to the features of the equipment.

[0087] The extraction unit 140 according to this embodiment groups elements of the candidate group into clusters according to the operation characteristics of each piece of equipment in the target building, and extracts equipment control setting candidates for each cluster. Specifically, when extracting equipment control settings, the extraction unit 140 defines each cluster by appropriately grouping equipment control setting candidates with the same or similar operation characteristics of corresponding equipment within the equipment controllable range based on the equipment characteristic data 431. The extraction unit 140 then extracts only equipment control settings that match the preferences of the equipment manager from each cluster. Note that multiple clusters are defined by the extraction unit 140. Specifically, each cluster is a set defined according to one of the following equipment operation characteristics (features), or a combination of these: Feature 1: The number of operating units of the corresponding heat sources is the same. When each cluster is defined according to Feature 1, the number of operating units of the corresponding heat sources differs between different clusters. Feature 2: The air conditioning temperature setting trends in the corresponding multiple rooms are the same. When each cluster is defined according to feature 2, the air conditioning temperature settings in corresponding rooms differ between different clusters. Feature 3: The trends in the air volume settings of corresponding fans are the same. When each cluster is defined according to feature 3, the fan air volume settings differ significantly between different clusters. Feature 4: The trends in the corresponding operating time periods are the same. When each cluster is defined according to feature 4, the corresponding start times and end times differ significantly between different clusters. The extraction unit 140 may use the learned inference model 432 as the inference model 131 before the inference model 131 is learned.

[0088] The data acquisition unit 190 according to this embodiment may acquire the trained inference model 432 .

[0089] The trained inference model 432 is a model that infers the preferences of other facility managers based on data acquired about other buildings. That is, the trained inference model 432 is a model that has learned the facility control setting preferences about other buildings and is a model for inferring the facility control setting preferences. The other building data 191 according to this embodiment may be the trained inference model 432.

[0090] ***Explanation of Operation*** The following describes the differences between the operation during inference of the equipment control device 100 according to this embodiment and the operation during inference of the equipment control device 100 according to the first embodiment.

[0091] (Step S102) This step is basically the same as step S102 according to embodiment 1. However, the extraction unit 140 defines each cluster according to the operation characteristics of the equipment based on the equipment characteristic data 431, and extracts only equipment control settings that match the preferences of the equipment manager in each cluster.

[0092] Fig. 12 is a diagram illustrating the processing of the extraction unit 140. Fig. 12(a) shows an equipment controllable range and all of the equipment control setting candidates. Fig. 12(b) is a diagram corresponding to Fig. 12(a) and shows each cluster and an example of the equipment control setting candidates extracted in each cluster.

[0093] ***Explanation of Effect of Fourth Embodiment*** In this embodiment, equipment control setting candidates are extracted and presented for each cluster, which groups together equipment control setting candidates having similar operation characteristics of the corresponding equipment. Therefore, according to this embodiment, an equipment manager can more easily select equipment control settings while taking into consideration the equipment operation characteristics required by the equipment manager.

[0094] ***Other Embodiments*** The above-described embodiments can be freely combined, or any of the components of each embodiment can be modified, or any of the components can be omitted from each embodiment. Furthermore, the embodiments are not limited to those shown in embodiments 1 to 4, and various modifications are possible as needed. The procedures described using flowcharts, etc., can be modified as appropriate.

[0095] 11 Processor, 12 Memory, 13 Auxiliary storage device, 14 Input / output IF, 15 Communication device, 18 Processing circuit, 19 Signal line, 20 Equipment management device, 30 Display device, 40 Input device, 90 Equipment control system, 100 Equipment control device, 110 Optimization calculation unit, 120 Memory unit, 130 Learning unit, 131 Inference model, 140 Extraction unit, 150 Display unit, 160 Input unit, 170 Selection unit, 180 Communication unit, 190 Data acquisition unit, 191 Other building data, 210 Data selection unit, 310 Data mixing unit, 320 Data extraction unit, 330 Data expansion unit, 391 Other building data, 431 Equipment characteristic data, 432 Learned inference model.

Claims

1. An equipment control device that supports decision-making regarding equipment control settings for a target building, comprising: a data acquisition unit that acquires other building data, which is data based on equipment control setting preferences for other buildings different from the target building; an optimization calculation unit that uses optimization technology to create multiple equipment control setting candidates as elements of a group of candidates to be created within the equipment controllable range for the target building; and an extraction unit that uses an inference model, which is a model based on the other building data and is a model that infers equipment control setting preferences, to extract one or more equipment control setting candidates from the group of candidates to be created as elements of a group of candidates to be extracted.

2. The equipment control device according to claim 1 further comprises: a display unit that displays each element of the group of extraction candidates to the equipment manager of the target building; a memory unit that records data linking the group of extraction candidates with elements selected by the equipment manager from the group of extraction candidates as learning data; and a learning unit that re-learns the inference model based on the learning data.

3. The other building data is data for learning the inference model and is data indicating each feature of the other building, and the equipment control device further comprises a data selection unit that refers to each feature of the other building and selects data from the other building data as selected data in accordance with each feature of the target building, and the learning unit learns the inference model based on the selected data. Equipment control device as described in claim 2.

4. The other building data is data for learning the inference model, and the equipment control device further includes a data mixing unit that generates mixed data by mixing at least a portion of the learning data with at least a portion of the other building data, and the learning unit learns the inference model based on the mixed data. Equipment control device according to claim 2.

5. The other building data is data indicating each characteristic of the other building, and the equipment control device further comprises a data extraction unit that extracts data matching each characteristic of the target building from the other building data as extracted building data, and the data mixing unit uses the extracted building data instead of the other building data.The equipment control device described in claim 4.

6. The equipment control device further includes a data expansion unit that generates data for training the inference model based on the other building data as extended learning data, and the data mixing unit generates the mixed data by mixing at least a portion of the training data, at least a portion of the other building data, and at least a portion of the extended learning data, as described in claim 4 or 5.

7. An equipment control device as described in any one of claims 1 to 6, wherein the extraction unit groups the elements of the candidate group into clusters according to the operating characteristics of each piece of equipment equipped in the target building, and extracts candidates for equipment control settings in each cluster.

8. The other building data is a learned inference model that is a model that has learned preferences for equipment control settings related to the other building and is a model for inferring preferences for equipment control settings, and the extraction unit uses the learned inference model as the inference model before the inference model is learned, an equipment control device described in any one of claims 1 to 7.

9. An equipment control method executed by an equipment control device, which is a computer that supports decision-making regarding equipment control settings for a target building, wherein the equipment control device acquires other building data, which is data based on equipment control setting preferences for other buildings different from the target building; the equipment control device uses optimization technology to create multiple equipment control setting candidates as elements of a group of candidates to be created within the equipment controllable range for the target building; and the equipment control device uses an inference model, which is a model based on the other building data and is a model that infers equipment control setting preferences, to extract one or more equipment control setting candidates from the group of candidates to be created as elements of a group of candidates to be extracted.

10. An equipment control program executed by an equipment control device, which is a computer that supports decision-making regarding equipment control settings for a target building, which causes the equipment control device to execute the following: a data acquisition process that acquires other building data, which is data based on equipment control setting preferences for other buildings different from the target building; an optimization calculation process that uses optimization technology to create multiple equipment control setting candidates as elements of a group of candidates to be created within the equipment controllable range for the target building; and an extraction process that uses an inference model, which is a model based on the other building data and is a model that infers equipment control setting preferences, to extract one or more equipment control setting candidates from the group of candidates to be created as elements of a group of candidates to be extracted.

Images