WO2023232473A1 - Commande prédictive d'un appareil électroménager - Google Patents

Commande prédictive d'un appareil électroménager Download PDF

Info

Publication number
WO2023232473A1
WO2023232473A1 PCT/EP2023/063252 EP2023063252W WO2023232473A1 WO 2023232473 A1 WO2023232473 A1 WO 2023232473A1 EP 2023063252 W EP2023063252 W EP 2023063252W WO 2023232473 A1 WO2023232473 A1 WO 2023232473A1
Authority
WO
WIPO (PCT)
Prior art keywords
component
household appliance
control messages
controlled
controlled component
Prior art date
Application number
PCT/EP2023/063252
Other languages
German (de)
English (en)
Inventor
Thomas Braune
Gudrun Schliecker
Simon Letzgus
Original Assignee
BSH Hausgeräte GmbH
Technische Universität Berlin
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 BSH Hausgeräte GmbH, Technische Universität Berlin filed Critical BSH Hausgeräte GmbH
Publication of WO2023232473A1 publication Critical patent/WO2023232473A1/fr

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B15/00—Systems controlled by a computer
    • G05B15/02—Systems controlled by a computer electric
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/20—Pc systems
    • G05B2219/26—Pc applications
    • G05B2219/2613—Household appliance in general

Definitions

  • the invention relates to the control of a household appliance.
  • the invention relates to the control of a functional component of a household appliance to reduce energy consumption.
  • a household appliance includes various functional components, each of which provides a predetermined function within the household appliance.
  • functional components of a washing machine include a pump, a drive motor or its control, or a heater or its control.
  • Functional components are connected to one another using one or more control buses. Using the control buses, the functional components coordinate with one another using instructions and data to carry out device functions.
  • All electrical functional components of a household appliance are supplied with electrical power as long as the household appliance itself is switched on or is completing a predetermined functional run. If a control message for executing a function of a functional component arrives, the requested function of the functional component is executed. Outside of the functional executions, the functional component remains switched on in order to be able to recognize a control message addressed to it. This means that the functional component can draw power even when it is not being used to operate the household appliance.
  • An activation or deactivation of the functional component within a functional run of the household appliance usually takes place depending on predetermined processes and current conditions.
  • a functional component is activated is usually not known by the functional component.
  • a control component for the functional component cannot support a temporary shutdown of the functional component.
  • a first method for controlling a household appliance with at least one controlled and one controlling component, between which control messages for controlling the household appliance are transmitted comprises steps of detecting control messages between the components; detecting activation of the controlled component; and machine learning a pattern of control messages that predicts subsequent activation of the controlled component.
  • Machine learning methods can therefore detect patterns of control messages that indicate impending activation of the controlled component.
  • a component can be trained which, after training, can be used to switch the controlled component on and/or off depending on control messages.
  • the component can be trained to recognize one or more predetermined message patterns.
  • a message pattern or pattern of control messages is formed here by a temporal sequence of control messages that are transmitted between components of the household appliance.
  • a chronological order or message content can vary within definable limits.
  • the aim of machine learning is to provide a switch-on signal prior to activation.
  • Supervised learning is preferably used. Nevertheless, it may not be necessary for a person to provide learning content; rather, a message pattern can be automatically determined and learned on a household appliance that performs its function.
  • input data and/or target variables for the activation of a controlled component can be collected from existing message protocols.
  • the determination of a switch-off signal for a component to be controlled can be trained or learned in a corresponding manner.
  • KRR Kernel Ridge Regression
  • KNN artificial neural network
  • a second method for controlling a household appliance having a first and at least a second component, between which control messages for controlling the household appliance are transmitted comprises steps of detecting control messages between the components; recognizing a pattern of control messages indicative of subsequent activation of the first component; and switching on the functional component before activation takes place.
  • the second method can correspond to parts of the first method, with the difference that a recognition learned in the first method is used in the second method to control a household appliance or to switch a first component of the household appliance on or off. While the first method is preferably implemented as part of the development or completion of the household appliance, the second method can be carried out as part of the use of the household appliance by a user, in particular in a household.
  • the first and second methods can each be carried out using a processing device that includes a programmable microcomputer or microcontroller.
  • the first method is usually carried out using an external processing device and the second using a processing device included in the household appliance. Both methods can each be in the form of a computer program product with program code means.
  • the computer program product can also be stored on a computer-readable data carrier.
  • the communication connection can be a direct connection, a data bus or a multi-component data bus.
  • Message patterns to be learned on a data bus can also include messages that are not directly exchanged between the controlling and controlled components.
  • a data bus as a means of exchanging messages between multiple participants also enables multiple combinations of controlling components.
  • a data bus with a defined, known or standardized message exchange enables subsequent to make the exchange of the participating components traceable using appropriate measurement technology.
  • the message exchange takes place independently of the resulting energy control, so that messages can be exchanged before and in the training phase independently of the energy control of the controlled component. This separation makes it possible to obtain target information about the activity of the controlled component, e.g. via power consumption measurement.
  • the energy supply of the controlled component is energetically separated from the controlled component. This means that the energy to the controlled component can be switched on and off at any time.
  • the energy supply unit is preferably set up to switch on the controlled component a predetermined period of time before its expected activation. In this way, the controlled component can be initialized or made ready before activation occurs. Optionally, there can also be a predetermined period of time between the end of activation and switching off in order to allow the controlled component to function following activation.
  • a processing device is provided which is set up to determine a pattern of control messages which indicates a phase of a predetermined length without activation of the controlled component; and turn off the controlled component when the pattern is detected.
  • the controlled component is only switched off after it has finished its activities.
  • the proper execution of the defined functions of the controlled component must always be ensured by the energy control.
  • control message is converted into a vector presentation using Natural Language Processing (NLP).
  • NLP Natural Language Processing
  • a control message includes one or more instructions, optionally supplemented by data.
  • the instruction and data are put into a semantic context that improves qualified message analysis.
  • other than active/inactive events may also be predicted based on patterns of control messages between components. For example, a time when the function is expected to be completed can be predicted. Other events such as the impending end of a functional phase that is included in the functional run can also be predicted. For example, the time when a washing machine switches from washing to spinning can be predicted in order to inform the customer.
  • the household appliance is preferably set up to carry out a predetermined function taking various parameters into account; wherein a time of expected completion of the function is predicted based on transmitted control messages.
  • Figure 1 shows a household appliance
  • Figure 2 shows a flow chart of a first method
  • Figure 3 shows a flowchart of a second method.
  • FIG. 1 shows a household appliance 100, which is shown purely as an example as a washing machine.
  • the household appliance 100 includes, for example, several controlled components 105 and a controlling component 110.
  • a controlled component 105 which is also referred to herein as the first component 105 or first functional component 105, is set up to fulfill a predetermined function within the household appliance 100. This function is usually specific or specialized and can be provided using a predetermined electrical load.
  • a controlled component 105 may include a motor, a heater, a fan, a detergent dispenser, a steam generator, an ozone source, a user interface, or an optical function indicator, such as a light ring.
  • the controlling component 110 which is also referred to herein as second component 110 or second functional component 110, is designed, for example, as a control component or control device.
  • the controlling component 110 can be set up to coordinate the interaction of controlled components 105.
  • the components 105, 110 are communicatively connected to one another by means of a data bus 115.
  • a component 105, 110 can send a control message to one or more other components 105, 110 on the data bus 115.
  • a control message can, for example, describe a state or request a function of another component 105, 110.
  • a power supply unit 120 is set up to supply components 105, 110 with electrical energy.
  • the energy supply unit 120 includes a processing device 135 and a series of switches 140, each of which is set up to switch at least one controlled component 105 on and off. To determine when a controlled component 105 must be switched on or off, the power supply unit 120 can be connected to the data bus 115 by means of an interface 130.
  • the processing device 135 is provided outside the energy supply unit 120.
  • a switch 140 can also be provided at another location, for example on or in an associated controlled component 105.
  • the processing device 135 is preferably set up to record control messages on the data bus 115 and to recognize a predetermined pattern of control messages, which in particular indicates an imminent activation or deactivation of a controlled component 105. It is further preferred that the controlled component 105 is switched on by means of an associated switch 140 when imminent activation has been detected. In a corresponding manner, a further pattern of control messages can also be recognized on the data bus 115, which indicates an imminent deactivation of a predetermined controlled component 105. In addition, it can be determined that a phase without activation of the controlled component 105 is impending. In this case, the controlled component 105 can be switched off using an associated switch 140.
  • a controlled component 105 is switched on a predetermined period of time before it is activated, so that the controlled component 105 is fully initialized or prepared at the time of its activation. It is also preferred that the controlled component 105 is switched off promptly after the end of its use. Depending on the component use, switching off can take place by generating a specially trained switch-off signal or after an estimated time has elapsed after which use has ended.
  • the recognition of a predetermined pattern of control messages and the resulting switching on or off of a controlled component 105 is carried out using machine learning techniques.
  • the processing device 135 in the household appliance 100 records either a learning system or the results of a message pattern learning system.
  • the subsystem establishes a connection between captured control messages and an activation or deactivation of the controlled component 105.
  • an activation of a controlled component 105 can be determined directly by the processing device 135 on the basis of a control message. Activation of a controlled component 105 can also be determined when a current drawn by it, which can be determined using an associated current sensor 145, increases above a predetermined threshold value. Deactivation of a controlled component 105 can be determined if a corresponding control message arrives or the current consumed by it drops below a further predetermined threshold value.
  • switches 140 are closed in order to be able to observe a learnable process of the household appliance 100. Based on observed control messages, it can be predicted when an activation of a predetermined controlled component 105 is to be expected. The prediction can be checked below based on the activation of the controlled component 105. This information can be sufficient to determine or automatically learn a suitable pattern of control messages.
  • a switch 140 can be closed or opened depending on an impending activation or a successful deactivation of a controlled component 105 in order to essentially only supply the controlled component 105 with power when it is also activated.
  • Machine learning can continue in parallel with the use of prediction to achieve improved discriminatory power for the corresponding pattern.
  • the processing device 135 is sufficiently trained to recognize one or more patterns of control messages on the data bus 115 and to make a prediction about activation or deactivation of a controlled component 105, the household appliance 100 can be delivered to an end customer. Learning can be done under laboratory conditions, for example by running through as many different variations of existing functional programs of the household appliance 100 as possible. This means that the trained processing device 135 can be used on a large number of identical or identical household appliances 100. In such a case, the current sensors 145 are not required in a delivered household appliance 100.
  • FIG. 2 shows a controlled method 200 for training a processing device 135 of a household appliance 100. If a processing device 135, which is later to be used to switch at least one controlled component 105 on and off, is not sufficiently powerful for the second method 200, then one can temporary or external processing device 135 can be inserted into the household appliance 100 or connected to it.
  • activation of a predetermined controlled component 105 may be determined.
  • an activation or deactivation message can be recorded on the data bus 115 or a current consumption of the controlled component 105 can be monitored by means of an assigned current sensor 145 to determine whether it exceeds or falls below a predetermined threshold value.
  • an activity of the controlled component 105 can be detected. For this purpose, either an activation message or a current consumption of the controlled component 105 that is above a predetermined threshold value can be detected. In a step 215 an end of the activity can be detected. The determination can be determined based on a detected message to deactivate the controlled component 105 or a current consumption of the component 105 that has fallen below a predetermined threshold.
  • a control message can be recorded on the data bus 115 in a step 220. If possible, all control messages on the data bus 115 are analyzed, and not just those that originate from or are directed to the predetermined controlled component 105. In other embodiments, only predetermined control messages whose transmitters or receivers include predetermined components 105, 110 can also be taken into account.
  • a received control message can be converted into a vector representation.
  • the control message can be sent as a sentence using NLP or linguistic statement can be understood and converted accordingly. This can result in an algorithmic representation of the control message or one that can be processed using adaptive techniques.
  • a correlation between control messages and activations or deactivations can be learned. Emphasis is placed on a temporal connection.
  • a pattern can be determined that precedes activation of the controlled component 105.
  • Another pattern may be determined that precedes impending deactivation of the controlled component 105. In both cases, it can also be determined what time interval is to be expected between the appearance of the pattern and the activation or deactivation of the controlled component 105.
  • a switch-on signal can be generated which lies at a predetermined time interval before the controlled component 105 is activated.
  • a switch-off signal can be generated which occurs at a predetermined time interval after the controlled component 105 has been deactivated.
  • the learning of the correlation in step 230 can be carried out in a self-reinforcing manner in such a way that the switch-on signal and/or the switch-off signal meets the time requirements regarding the activation or deactivation of the controlled component 105 with ever-improving accuracy.
  • the learning can be considered complete if the activation or deactivation of the controlled component 105 does not deviate from the actual event by more than a predetermined amount of time with a predetermined probability.
  • FIG 3 shows a flowchart of a second method 300 for controlling a household appliance 100.
  • the second method 300 is similar to the controlled method 200, but is usually carried out by an internal processing device 135 of the household appliance 100.
  • a control message is recorded on the data bus 115.
  • This step essentially corresponds to step 220 of the controlled method 200.
  • a captured control message can be converted into a vector representation.
  • This step essentially corresponds to step 225 of the controlled method 200.
  • a pattern in the captured control messages may be recognized that indicates imminent activation of a predetermined controlled component 105.
  • the pattern can be recognized in particular by means of a processing device 135, which is trained for this purpose according to a controlled method 200 described herein. Based on the recognized pattern, a switch-on signal can be generated that precedes activation by a predetermined time period.
  • the controlled component 105 can be switched on depending on the switch-on signal.
  • a pattern can also be recognized that indicates an impending deactivation or an impending phase of inactivity of the controlled component 105.
  • a switch-off signal can be generated that is as close as possible to the start of inactivity of the controlled component 105. Based on this signal, the controlled component 105 can be switched off in a step 325. It is preferred to avoid switching off a controlled component 105 that has not yet been deactivated.
  • control component 110 controlling component, control component

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Selective Calling Equipment (AREA)

Abstract

L'invention concerne un procédé (300) de commande d'un appareil électroménager (100) à l'aide d'au moins un composant commandé (105) et d'un composant de commande (110), entre lesquels des messages de commande sont transmis afin de commander l'appareil électroménager (100). Le procédé comprend les étapes consistant à détecter (305) des messages de commande entre les composants (105, 110) ; identifier (315) un motif de messages de commande, ledit motif indiquant une activation ultérieure du composant commandé (105) ; et mettre sous tension (320) le composant fonctionnel avant que le processus d'activation ne soit effectué.
PCT/EP2023/063252 2022-06-03 2023-05-17 Commande prédictive d'un appareil électroménager WO2023232473A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102022205687.6 2022-06-03
DE102022205687.6A DE102022205687B3 (de) 2022-06-03 2022-06-03 Vorhersagendes Steuern eines Haushaltsgeräts

Publications (1)

Publication Number Publication Date
WO2023232473A1 true WO2023232473A1 (fr) 2023-12-07

Family

ID=86558743

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2023/063252 WO2023232473A1 (fr) 2022-06-03 2023-05-17 Commande prédictive d'un appareil électroménager

Country Status (2)

Country Link
DE (1) DE102022205687B3 (fr)
WO (1) WO2023232473A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160274611A1 (en) * 2015-03-17 2016-09-22 Cielo WiGle Inc. Smart electrical switch with engery metering capability
WO2018033715A1 (fr) * 2016-08-15 2018-02-22 Nambiar Krishnan Ratnakaran Dispositif amélioré de commande
US20180239311A1 (en) * 2015-02-24 2018-08-23 Energy Technologies Institute Llp Method and Apparatus for Controlling an Environment Management System within a Building
US20190353379A1 (en) * 2018-05-16 2019-11-21 Johnson Controls Technology Company Building management hvac control using human sensors
US20200220789A1 (en) * 2002-06-18 2020-07-09 Apple Inc. Learning device interaction rules

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160049789A1 (en) 2013-04-11 2016-02-18 Liricco Technologies Ltd. Energy management system
KR20180034955A (ko) 2016-09-28 2018-04-05 엘지전자 주식회사 전자기기 및 전자기기의 제어방법

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200220789A1 (en) * 2002-06-18 2020-07-09 Apple Inc. Learning device interaction rules
US20180239311A1 (en) * 2015-02-24 2018-08-23 Energy Technologies Institute Llp Method and Apparatus for Controlling an Environment Management System within a Building
US20160274611A1 (en) * 2015-03-17 2016-09-22 Cielo WiGle Inc. Smart electrical switch with engery metering capability
WO2018033715A1 (fr) * 2016-08-15 2018-02-22 Nambiar Krishnan Ratnakaran Dispositif amélioré de commande
US20190353379A1 (en) * 2018-05-16 2019-11-21 Johnson Controls Technology Company Building management hvac control using human sensors

Also Published As

Publication number Publication date
DE102022205687B3 (de) 2023-10-12

Similar Documents

Publication Publication Date Title
DE102006029708B4 (de) Verfahren und Vorrichtung zur Steuerung der Standby-Leistung einer Waschmaschine
DE602005004997T2 (de) Diagnostisches Verfahren und System
DE19880536B4 (de) Visuelles Programmierverfahren und dieses Verfahren anwendendes Pogrammiersystem
DE60030758T2 (de) Leichtes schienenfahrzeug mit einem voraussagenden diagnosesystem für motorgetriebene automatische türen
DE102018111892B4 (de) Betriebsüberwachungsvorrichtung und Steuerprogramm dafür
DE2648229A1 (de) Einschaltkreis als urlader fuer digitalrechner
DE102017107284A1 (de) Verfahren und steuergerät zum überwachen eines bordnetzes eines fahrzeugs
EP3704573B1 (fr) Procédé pour la réalisation d'une mise à jour de logiciel dans un appareil de commande d'un véhicule à moteur ainsi que véhicule à moteur ainsi aménagé
DE102018006992B4 (de) Steuereinrichtung
EP3781922A1 (fr) Système de diagnostic et procédé de traitement de données d'un véhicule à moteur
DE102018000579A1 (de) Überwachen einer Funktionsbereitschaft eines elektrischen Gerätes
EP3179372A1 (fr) Procédé et dispositif de test d'une pluralité d'unités de commande d'une unité technique
DE102022205687B3 (de) Vorhersagendes Steuern eines Haushaltsgeräts
WO2022117622A1 (fr) Procédé et dispositif de détermination de durée de vie restante de système technique
DE102017000477A1 (de) Leiterprogrammanzeigevorrichtung mit automatischer Verfolgungsfunktion für eine Selbsthalteschaltung des Leiterprogramms
WO2013131556A1 (fr) Procédé et dispositif de commande économisant l'énergie d'un équipement
EP3467600A1 (fr) Module central de nuage pour une commande programmable par mémoire
WO2016146506A1 (fr) Appareil d'établissement de projet et procédé pour configurer et/ou paramétrer des composants d'automatisation d'un système d'automatisation
EP2971769B1 (fr) Entrée de schéma r&i pour un procédé de contrôle et/ou de surveillance d'un système de compresseurs
WO2012119680A1 (fr) Procédé d'optimisation des temps morts dans le déroulement de la production par des machines-outils
EP2712429A1 (fr) Commande d'un système technique
DE102019213001A1 (de) Wissensproduktionssystem
WO2022238230A1 (fr) Procédé mis en œuvre par ordinateur pour fournir un modèle d'apprentissage automatique entraîné pour détecter au moins un événement relatif à un système technique
DE102010022141A1 (de) Verfahren und Assistenzsystem zur Unterstützung des Bedieners eines Computers
EP3018542B1 (fr) Installation d'automatisation ayant un mode pause à réduction de puissance

Legal Events

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

Ref document number: 23726137

Country of ref document: EP

Kind code of ref document: A1