WO2024034110A1 - Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande - Google Patents

Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande Download PDF

Info

Publication number
WO2024034110A1
WO2024034110A1 PCT/JP2022/030704 JP2022030704W WO2024034110A1 WO 2024034110 A1 WO2024034110 A1 WO 2024034110A1 JP 2022030704 W JP2022030704 W JP 2022030704W WO 2024034110 A1 WO2024034110 A1 WO 2024034110A1
Authority
WO
WIPO (PCT)
Prior art keywords
controlled system
actual
model
identified
conditions
Prior art date
Application number
PCT/JP2022/030704
Other languages
English (en)
Japanese (ja)
Inventor
寛光 穗苅
護 土井
Original Assignee
三菱電機株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 三菱電機株式会社 filed Critical 三菱電機株式会社
Priority to PCT/JP2022/030704 priority Critical patent/WO2024034110A1/fr
Priority to JP2022577235A priority patent/JPWO2024034110A1/ja
Publication of WO2024034110A1 publication Critical patent/WO2024034110A1/fr

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/50—Air quality properties
    • F24F2110/64—Airborne particle content

Definitions

  • the present disclosure relates to a control device and a control method.
  • the air conditioning control device described in Patent Document 1 uses state information regarding air conditioning and a plurality of air conditioning models (for example, an energy model, a statistical model, and a thermal fluid simulation model) for estimating the indoor temperature at a target indoor position.
  • the operation of the air conditioner is controlled based on the indoor temperature at the target position in the room, which is estimated using either of the following.
  • this air conditioning control device which of the plurality of air conditioning models to use is set in advance for each time period or day of the week, for example.
  • the external conditions or internal conditions of the controlled system described above can change from moment to moment. Therefore, as the model for the controlled system, it is preferable to select an appropriate model according to the above conditions.
  • a model related to the controlled system to be used is set in advance, and the control device controls the operation of the controlled system using a model suitable for the operating conditions of the controlled system that change from moment to moment. The problem was that it didn't exist.
  • An object of the present disclosure is to provide a control device, a chiller system, a simulation section, an actual device section, and a control method that can control the operation of a controlled system using a model suitable for the operating conditions of the controlled system. It is in.
  • a control device provides a plurality of assumed conditions and a plurality of settings to be set in the controlled system under the plurality of conditions in order to control the operation of the controlled system.
  • a simulator unit that generates a trained model having a plurality of identified models whose relationships with set values have been identified; and a simulator unit that generates a trained model having a plurality of identified models whose relationships with set values are identified; and an actual device section that sets actual setting values corresponding to one of the identified models to the controlled system.
  • the operation of the controlled system can be controlled using a model suitable for the operating conditions of the controlled system.
  • FIG. 7 illustrates the generation of an identified model g ⁇ of an embodiment
  • control device SD of the embodiment will be explained.
  • a single code may be used to collectively refer to multiple names.
  • “identified model g ⁇ ” may be replaced by "identified model g ⁇ 1 , g ⁇ ⁇ 2 , g ⁇ 3 3, ,,,'' are sometimes collectively called.
  • FIG. 1 is a functional block diagram of the control device SD and the target system TS of the embodiment.
  • the functions of the control device SD and the target system TS of the embodiment will be described with reference to FIG. 1.
  • the control device SD of the embodiment controls the operation of the target system TS, as shown in FIG.
  • the target system TS includes a control system SS and a controlled system HS.
  • the control system SS and the controlled system HS may each be arbitrary systems.
  • the controlled system HS may be, for example, a chiller system.
  • the control system SS is, for example, a building management system that manages a building in which the chiller system is installed.
  • control device SD includes a simulator section SM and a real machine section JK.
  • the simulator unit SM determines the relationship between a plurality of assumed conditions ⁇ and a plurality of setting values c to be set in the controlled system HS under the plurality of conditions ⁇ for controlling the operation of the controlled system HS.
  • a trained model f (for example, a neural network) having a plurality of identified models g is generated.
  • the simulator unit SM is a trained model having a plurality of identified models g ⁇ , that is, identified models g ⁇ 1 , g ⁇ 2 , g ⁇ 3 , , , for controlling the operation of the controlled system HS. Generate a model f ⁇ . Each of the identified models g ⁇ 1 , g ⁇ 2 , g ⁇ 3 , . relationships have been identified.
  • the actual machine part JK identifies one of the plurality of identified models g ⁇ 1 , g ⁇ 2 , g ⁇ 3 , , , which is specified by inputting the actual condition ⁇ of the controlled system HS into the learned model.
  • the actual setting values corresponding to the completed model are set in the controlled system HS.
  • the actual machine unit JK obtains the set value c ⁇ to be set in the controlled system HS by referring to the learned model f ⁇ .
  • the actual machine section JK also derives a control amount u ⁇ for controlling the operation of the controlled system HS by referring to the learned model f ⁇ based on the actual condition ⁇ of the controlled system HS.
  • FIG. 2 shows the hardware configuration of the control device SD of the embodiment.
  • control device SD includes a processing circuit SH, and further includes an input circuit NY and an output circuit SY as necessary.
  • the processing circuit SH is dedicated hardware.
  • the processing circuit SH realizes the functions of the simulator section SM (shown in FIG. 1) and the actual machine section JK (shown in FIG. 1) of the control device SD.
  • the processing circuit SH is, 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 input circuit NY and the output circuit SY exchange inputs and outputs related to the operation of the processing circuit SH with the outside of the control device SD.
  • FIG. 3 shows a hardware configuration based on software implementation of the control device SD of the embodiment.
  • control device SD includes a processor PR and a memory circuit KI, and further includes an input circuit NY and an output circuit SY as necessary.
  • the processor PR is a CPU (also referred to as a Central Processing Unit, central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, or DSP (Digital Signal Processing)) that executes a program.
  • the processor PR realizes the functions of the simulator section SM (shown in FIG. 1) and the actual machine section JK (shown in FIG. 1) of the control device SD.
  • the processor PR realizes the above functions using software, firmware, or a combination of software and firmware.
  • Software and firmware are written as programs and stored in the storage circuit KI.
  • the processor PR realizes the above functions by reading out and executing the above program from the storage circuit KI. It can be said that the above-mentioned program causes a computer to execute the procedures and methods of the simulator section SM and the actual machine section JK.
  • the memory circuit KI is a nonvolatile memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), etc.
  • RAM Random Access Memory
  • ROM Read Only Memory
  • flash memory EPROM (Erasable Programmable Read Only Memory)
  • EEPROM Electrically Erasable Programmable Read-Only Memory
  • These include flexible or volatile semiconductor memories, magnetic disks, flexible disks, optical disks, compact disks, mini disks, DVDs (Digital Versatile Discs), and the like.
  • the functions of the simulator section SM and the actual device section JK of the control device SD can be realized by hardware, software, firmware, or a combination thereof.
  • the input unit NY and the output unit SY exchange inputs and outputs related to the operation of the processor PR with the outside of the control device SD.
  • FIG. 4 is a flowchart showing the operation of the simulator unit SM of the embodiment.
  • FIG. 5 shows the operations of generation of learning data GD, generation of learned model f ⁇ , and optimization by the simulator unit SM of the embodiment.
  • control device SD of the embodiment will be explained with reference to FIGS. 4 and 5.
  • learning data GD is data for generating a trained model f ⁇ , and refers to a combination of condition ⁇ , setting value c, and simulation data D.
  • Step ST11 The simulator unit SM (shown in FIG. 1) sets conditions ⁇ in order to generate learning data GD through simulation. More specifically, the simulator unit SM sets the condition ⁇ within the assumption that the controlled system HS operates, and sets, for example, the condition ⁇ 1, the condition ⁇ 2, the condition ⁇ 3, etc., as shown in FIG. .
  • condition ⁇ is a condition under which the controlled system HS operates (operating condition of the controlled system).
  • the condition ⁇ is, for example, at least one of an external factor and a target value.
  • External factors are factors existing outside the controlled system HS that affect the operation of the controlled system HS.
  • the target value is a target value that the operation of the controlled system HS should achieve.
  • Condition ⁇ may include, for example, a plurality of external factors, a plurality of target values, or both an external factor and a target value.
  • the external factor is, for example, the above-described outside temperature of the building
  • the target value is, for example, the peak value of power consumption of the above-described chiller system.
  • the setting of the condition ⁇ by the simulator unit SM is performed, for example, in the following order.
  • the designer of the control device SD defines a plurality of setting items and the possible widths of the items. For example, the designer defines the outside temperature during cooling to be 20° C. to 40° C., and defines the internally generated heat load to be 0 W to 4000 W.
  • the designer defines representative values for the plurality of items described above. For example, the designer defines representative values [20, 25, 30, 35, 40] for the maximum outside temperature, and representative values [0, 1000, 2000, 3000, 4000] for the average internally generated heat load. Define.
  • the designer uses simulation software to perform calculations for each of the above conditions ⁇ , and generates simulation data D.
  • noise noise that follows a uniform distribution or normal distribution, has an average of 0, and has a small variance
  • noise noise that follows a uniform distribution or normal distribution, has an average of 0, and has a small variance
  • Step ST12 The simulator unit SM determines a set value c ⁇ . More specifically, as shown in FIG. 5, the simulator unit SM determines, for example, a set value c1 ⁇ 1 , a set value c2 ⁇ 1 , a set value c3 ⁇ 1 , . . . for the condition ⁇ 1, and similarly, for the condition ⁇ 2 , the set value c1 ⁇ 2 , the set value c2 ⁇ 2 , the set value c3 ⁇ 2 , . . . are determined. The simulator unit SM similarly determines setting values c1 ⁇ 3 , etc. for other conditions ⁇ 3, etc.
  • the set value c ⁇ is a value set in the controlled system HS to define the configuration of the controlled system HS.
  • the set value c ⁇ is, for example, the number of chillers (or heat storage tanks) that should be operated among the plurality of chillers (or heat storage tanks) that constitute the above-mentioned chiller system. , what temperature should the water temperature of the chiller be set to, or when (at what time) should the air conditioner in the building mentioned above start operating?
  • the setting value c ⁇ determined by the simulator unit SM may be, for example, any value, and may be a default value of the simulator software, for example. This is because no matter what value the set value c ⁇ is, the set value c ⁇ is updated using an optimization method such as Bayesian optimization so that the set value c ⁇ becomes better.
  • the simulator part SM sets the set value c ⁇ to, for example, the above-described temperature. Decide how many chillers to run.
  • Step ST13 The simulator unit SM calculates simulation data D ⁇ using the condition ⁇ and the set value c ⁇ .
  • the simulation data D ⁇ is, for example, general time series data, and is, for example, a CSV format file in which the columns are state values such as the control amount u and the room temperature of equipment and the room, and the rows are times.
  • the simulator unit SM When given the condition ⁇ and the set value c ⁇ , the simulator unit SM performs a simulation for a certain period of time (for example, one day) based on the condition ⁇ .
  • the simulator unit SM outputs simulation data D ⁇ as a result of the simulation.
  • the simulator unit SM When the controlled system HS is a chiller system, the simulator unit SM performs the above calculation, that is, the simulation, using, for example, conventionally known building simulation software. More specifically, as shown in FIG. 5, the simulator unit SM calculates simulation data D1 ⁇ 1 from the condition ⁇ 1 and the set value c1 ⁇ 1 through the above-described simulation, and calculates the simulation data D1 ⁇ 1 from the condition ⁇ 1 and the set value c2 ⁇ 1 . D2 ⁇ 1 is calculated, and similarly, for example, simulation data D1 ⁇ 2 is calculated from the condition ⁇ 2 and the set value c1 ⁇ 2 . The simulator unit SM similarly calculates simulation data D1 ⁇ 3 and the like for other conditions ⁇ 3 and the like.
  • the simulator unit SM completes the generation of the learning data GD through steps ST11 to ST13 described above. Specifically, the simulator unit SM changes the condition ⁇ within the expected range, changes the set value c within the expected range, and then performs steps ST11 to ST13 to generate the learning data GD. Finish generating. As shown in FIG. 5, the learning data GD includes a condition ⁇ , a set value c ⁇ , and simulation data D ⁇ .
  • Step ST14 The simulator unit SM generates an identified model g ⁇ in order to generate a learned model f ⁇ . Specifically, as shown in FIG. 5, the simulator unit SM calculates, for example, the set value c1 ⁇ 1 and the simulation data D1 ⁇ 1 , the set value c2 ⁇ 1 and the simulation data D2 ⁇ 1 , and the set value c3 ⁇ 1 and the simulation data D3 for the condition ⁇ 1 .
  • an identified model g ⁇ 1 is generated, and similarly, for example, for condition ⁇ 2, set values c1 ⁇ 2 and simulation data D1 ⁇ 2 , the set value c2 ⁇ 2 and the simulation data D2 ⁇ 2 , the set value c3 ⁇ 2 and the simulation data D3 ⁇ 2 , . . . , the identified model g ⁇ 2 is generated by performing system identification.
  • the simulator unit SM similarly generates identified models g ⁇ 3 and the like for other conditions ⁇ 3 and the like.
  • identification using non-linear approximation may also be used.
  • the least squares method or the gradient descent method may be used.
  • FIG. 7 shows the generation of the identified model g ⁇ of the embodiment.
  • the identified model g ⁇ is generated, for example, by the following procedure.
  • the system identified model associated with the set value cj ⁇ will be expressed as g ⁇ _cj.
  • the condition ⁇ and the set value c1 ⁇ are fixed, and the system identified model g ⁇ _c1 is determined from the obtained simulation data D1 ⁇ .
  • the least squares method is used for system identification, and linear approximation of the state equation is performed.
  • Control calculation is performed for the set value c1 ⁇ and the identified model g ⁇ , and the evaluation value y ⁇ _c1 is obtained as an output.
  • Step ST15 The simulator unit SM generates a learned model f ⁇ .
  • the simulator unit SM generates a learned model f ⁇ that is composed of an identified model g ⁇ 1 , an identified model g ⁇ 2 , an identified model g ⁇ 3 , .
  • the trained model f ⁇ is a neural network model that outputs a set value c ⁇ specified by the condition ⁇ and an identified model g ⁇ when the condition ⁇ is input. It may also be constructed using a random forest or gradient boosting tree model instead of a neural network.
  • the simulator unit SM completes the generation of the trained model f ⁇ through steps ST14 and ST15 described above.
  • Step ST16 The simulator unit SM optimizes the setting value c ⁇ , the output y ⁇ , etc. For example, when given the condition ⁇ x , the simulator unit SM obtains the optimal setting value c x , output y x , and control amount u x using the learned model f ⁇ .
  • control amount u is a value set in the controlled system HS in order to control the operation of the controlled system HS.
  • the simulator unit SM obtains the output y1 x1 and the control amount u1 x1 from the identified model g ⁇ 1 by, for example, giving the input c1 x1 to the learned model f. , Similarly, by giving input c1 x2 to learned model f, output y1 x2 and control amount u1 x2 are obtained from identified model g ⁇ 2. Similarly, the simulator unit SM obtains outputs y3 x3 , etc., and control amounts u x3 , etc. (not shown) by providing other inputs c1 x3 (not shown), etc.
  • the output y is optimal means that the output is the most desirable value within the range of possible values of the output y.
  • the control amount u is, for example, the air volume of the fan of the air conditioner or the water temperature of the coil.
  • FIG. 6 is a flowchart showing the operation of the actual machine section JK of the embodiment.
  • the operation of the actual machine part JK of the embodiment will be explained with reference to the flowchart of FIG. 6.
  • Step ST21 The actual machine unit JK (shown in FIG. 1) inputs the actual condition ⁇ , for example, the condition ⁇ , from the controlled system HS to the learned model f ⁇ .
  • the input of the actual condition ⁇ may be performed, for example, mainly at the beginning of the day for the chiller system, or may be performed mainly every other day.
  • Step ST22 As suggested in FIG. 5 ("optimization" and “generate trained model f ⁇ " part), the actual machine part JK uses the trained model f ⁇ , which is a neural network, etc., from the condition ⁇ . The specified setting value c ⁇ and the identified model g ⁇ are obtained.
  • the trained model f ⁇ which is a neural network or the like, is configured to output the set value c ⁇ specified by the condition ⁇ and the identified model g ⁇ upon receiving the input of the condition ⁇ .
  • Step ST23 The actual machine unit JK acquires the control amount u ⁇ specified from the identified model g ⁇ , as suggested in FIG. 5 (“Optimization” and “Generate learned model f ⁇ ” part) do.
  • the actual machine section JK sets the set value c ⁇ to the target system TS, and also sets the operation of the target system TS based on the control amount u ⁇ , that is, sets the control amount u ⁇ itself to the target system TS.
  • the air volume of the air conditioner is controlled as the control amount u.
  • the condition ⁇ is selected among the plurality of identified models g ⁇ included in the learned model f ⁇ depending on the situation of the target system TS, that is, the condition ⁇ .
  • a suitable identified model g ⁇ can be used to control the operation of the target system TS.
  • the chiller system is an example of the controlled system HS, and the controlled system HS may be a system other than the chiller system.
  • a chiller system which is an example of the controlled system HS, may include the above-mentioned control device SD.
  • control device can be used to control the operation of the controlled system using a model suitable for the conditions under which the controlled system operates.

Landscapes

  • Feedback Control In General (AREA)

Abstract

Un dispositif de commande (SD) comprend : une unité de simulateur (SM) qui génère un modèle entraîné (fθ) qui est destiné à commander le fonctionnement d'un système commandé (HS) et comprend une pluralité de modèles à identification achevée (gθ) dans lesquels les relations entre une pluralité de conditions supposées (θ) et une pluralité de valeurs de réglage (cθ) sur lesquelles le système commandé (HS) doit être réglé sous la pluralité de conditions (θ) ont été identifiées ; et une unité de machine réelle (JK) qui règle le système commandé (HS) à une valeur de réglage réelle (cθα) qui correspond à un modèle à identification achevée (gθα) parmi la pluralité de modèles à identification achevée (gθ) qui est identifié par l'entrée d'une condition réelle (θα) du système commandé (HS) dans le modèle entraîné (fθ).
PCT/JP2022/030704 2022-08-12 2022-08-12 Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande WO2024034110A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/JP2022/030704 WO2024034110A1 (fr) 2022-08-12 2022-08-12 Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande
JP2022577235A JPWO2024034110A1 (fr) 2022-08-12 2022-08-12

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2022/030704 WO2024034110A1 (fr) 2022-08-12 2022-08-12 Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande

Publications (1)

Publication Number Publication Date
WO2024034110A1 true WO2024034110A1 (fr) 2024-02-15

Family

ID=89851252

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2022/030704 WO2024034110A1 (fr) 2022-08-12 2022-08-12 Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande

Country Status (2)

Country Link
JP (1) JPWO2024034110A1 (fr)
WO (1) WO2024034110A1 (fr)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2019066135A (ja) * 2017-10-04 2019-04-25 ファナック株式会社 空調制御システム
WO2022101989A1 (fr) * 2020-11-10 2022-05-19 三菱電機株式会社 Dispositif de climatisation et dispositif d'apprentissage du dispositif de climatisation

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2019066135A (ja) * 2017-10-04 2019-04-25 ファナック株式会社 空調制御システム
WO2022101989A1 (fr) * 2020-11-10 2022-05-19 三菱電機株式会社 Dispositif de climatisation et dispositif d'apprentissage du dispositif de climatisation

Also Published As

Publication number Publication date
JPWO2024034110A1 (fr) 2024-02-15

Similar Documents

Publication Publication Date Title
US11965666B2 (en) Control method for air conditioner, and device for air conditioner and storage medium
US6540148B1 (en) Method and apparatus for sequencing multistage systems of known relative capacities
US11733662B2 (en) Heuristic method of automated and learning control, and building automation systems thereof
CN112577161B (zh) 空调能耗模型训练方法与空调系统控制方法
US11549710B2 (en) Model predictive control-based building climate controller incorporating humidity
KR102198817B1 (ko) 열·공조 시스템에 대한 수요 반응 결정 모델을 생성하는 방법 및 수요 반응을 수행하기 위한 방법
US9429921B2 (en) Method and system for energy control management
US11042924B2 (en) Building controller for optimizing equipment upgrades with design of experiments
US20210254851A1 (en) Air conditioning control device and air conditioning control method
US10061274B2 (en) Heating, ventilation, and air conditioning system controller
Wang et al. Event-driven optimization of complex HVAC systems
Maasoumy et al. Selecting building predictive control based on model uncertainty
US20040186599A1 (en) System and method for model-based control of a building fluid distribution system
US20140039844A1 (en) Determining a heating, ventilation, and air conditioning model for a building
Kaam et al. Time-averaged ventilation for optimized control of variable-air-volume systems
KR20210089458A (ko) 빌딩 설비 에너지 관리 제어 시스템 및 그 제어 방법
Sporr et al. Automated HVAC control creation based on building information modeling (BIM): Ventilation system
US20190338978A1 (en) Operating an hvac system based on predicted indoor air temperature
Lara et al. Modeling and identification of the cooling dynamics of a tropical island hotel
KR20210047554A (ko) 건물 에너지 시뮬레이션을 위한 디지털 트윈 시스템 및 건물 에너지 시뮬레이션 방법
CN115167562B (zh) 一种机房温度控制方法和装置
WO2024034110A1 (fr) Dispositif de commande, système de refroidisseur, unité de simulation, unité de machine réelle et procédé de commande
Stephens The impacts of duct design on life cycle costs of central residential heating and air-conditioning systems
US10900683B2 (en) Heating, ventilation, and air conditioning system controller
Jorissen et al. Towards an automated tool chain for MPC in multi-zone buildings

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2022577235

Country of ref document: JP

Kind code of ref document: A

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22955025

Country of ref document: EP

Kind code of ref document: A1