WO2023193934A1 - Procédé et système de prise en charge de réduction d'émissions - Google Patents

Procédé et système de prise en charge de réduction d'émissions Download PDF

Info

Publication number
WO2023193934A1
WO2023193934A1 PCT/EP2022/067931 EP2022067931W WO2023193934A1 WO 2023193934 A1 WO2023193934 A1 WO 2023193934A1 EP 2022067931 W EP2022067931 W EP 2022067931W WO 2023193934 A1 WO2023193934 A1 WO 2023193934A1
Authority
WO
WIPO (PCT)
Prior art keywords
activities
sub
activity
knowledge graph
level
Prior art date
Application number
PCT/EP2022/067931
Other languages
English (en)
Inventor
Timo SZTYLER
Julia Gastinger
Original Assignee
NEC Laboratories Europe GmbH
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 NEC Laboratories Europe GmbH filed Critical NEC Laboratories Europe GmbH
Publication of WO2023193934A1 publication Critical patent/WO2023193934A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G06N5/022—Knowledge engineering; Knowledge acquisition
    • G06N5/025—Extracting rules from data

Definitions

  • the present invention relates to a computer-implemented method and a system for supporting emission reduction.
  • a computer-implemented method for supporting emission reduction comprising:
  • a system for supporting emission reduction comprising:
  • - a domain of interest module that is configured to receive sensor data from multiple sensors of a sensor network, to transform the sensor data into a knowledge graph, and to infer, from the knowledge graph, sub-activities that caused the sensor data, wherein the inferred sub-activities are added to the knowledge graph;
  • an isolation module that is configured to infer, from the knowledge graph, knowledge information about which sub-activities form a higher-level activity, wherein the inferred knowledge information is added to the knowledge graph such that isolated higher-level activities are identified in the knowledge graph;
  • an activity matching module that is configured to determine, in a historical database, matching candidates for the higher-level activities, wherein the matching candidates are added to the knowledge graph;
  • an emission estimation module that is configured to determine emission information for each sub-activity and each higher-level activity based on the matching candidates and historical data, wherein the determined emission information is added to the knowledge graph.
  • embodiments of the present invention provides a short-term approach that can be implemented right away. As such, the presented solution is not comparable with mid- and longterm solutions as these approaches are different and lengthy.
  • the present invention provides methods and systems for supporting emission reduction.
  • the invention provides a method and a system that can recognize a process or activity of an individual or company. Further, subactivities (i.e., for example, actions or intermediate steps) of that (higher-level) activity or process are analyzed to find alternative execution options. The goal is to substitute sub-activities such as intermediate steps in a way that the overall activity or process stays the same but the emissions or costs are reduced. This enables on the one hand to teach or educate individuals and on the other hand to optimize processes or decision making in a (industrial) company.
  • an embodiment of the invention provides a computer-implemented method for supporting emission reduction, wherein sensor data is collected from multiple sensors of a sensor network.
  • the sensor data is then transformed into a knowledge graph, such that the collected information can be analyzed in a suitable way.
  • Sub-activities, which caused the collected sensor data are inferred from the knowledge graph, in particular by the use of a machine learning model.
  • the inferred sub-activities are added to the knowledge graph.
  • knowledge information about which sub-activities form a higher-level activity is inferred from the knowledge graph and is also added to the knowledge graph, such that isolated higher-level activities are identified in the knowledge graph.
  • Matching candidates for the higher-level activities are determined in a historical database and are added to the knowledge graph. Finally, emission information is determined for each sub-activity and each higher-level activity based on the matching candidates and historical data. This determined emission information is added to the knowledge graph.
  • an emission reduction process is accomplishable, where sub-activities (such as actions or intermediate steps) of an performed activity or executed process may be automatically substituted with alternatives, such that the overall activity or goal stays the same but emissions are reduced.
  • sub-activity and “higher-level activity” may refer in particular in the claims, preferably in the description, to a structure that reflects a hierarchical concept where actions (higher-level activity) may consist of several - preferably atomic and/or intermediate - steps (sub-activity).
  • actions higher-level activity
  • sub-activity may consist of several - preferably atomic and/or intermediate - steps (sub-activity).
  • the same action or the same type of action can be realized in different ways, i.e. , may consist of different atomic/intermediate steps.
  • matching candidate may refer in particular in the claims, preferably in the description, to a matching candidate that is searched for a predetermined activity, wherein the match represents the same activity but the actual execution can be different to the execution of the predetermined activity.
  • the method may enable emission reduction in the context of controlling building energy performance, wherein the sensor network of a building may provide the sensor data.
  • the sensor network of a building may provide the sensor data.
  • Large companies, public buildings, or smart homes have to optimize the comfort of each thermal zone by taking into account individual thermodynamics of thermal zones and different weather impacts on the external surfaces or faces of each zone of a building.
  • the embodiment of the invention can be used to improve and determine the overall optimal building energy performance. This in turn may lead to reduced (carbon) emissions.
  • the sensor data may comprise thermal information about one or more thermal zones of the building and/or information about conditions of controllable entities of the building.
  • the information about conditions of controllable entities of the building may include conditions of doors, windows, number of people, and available devices such as air conditioning systems.
  • a desired state of a thermal zone of the building is considered as a higher-level activity that consists of several sub-activities.
  • a method and/or a system in accordance with an embodiment of the invention may help by reaching the desired state of each thermal zone.
  • the desired state of a thermal zone can be considered as an activity that consists of several sub-activities/actions such as opening the window for five minutes, reducing the cooling level of the air conditioning for that time span, and/or lowering the heat after people left the room.
  • the embodiment of the invention may select the one which causes as few (carbon) emissions as possible.
  • a list of (ranked) sub-activities/actions that should be performed to reach the desired thermal zones can be directly feed to the predictive building control system.
  • output of the method and/or system in accordance with an embodiment of the invention can be directly used as input to a predictive building control system, which is able to control, e.g., windows, doors, heating systems, air condition systems, etc.
  • the sensor data is represented in form of triples in the knowledge graph.
  • the knowledge graph can be implemented as a formalism to represent structured data, i.e. to organize, manage, and retrieve structured information.
  • the nodes of a knowledge graph may represent real-world entities along with their types and attributes, while the edges of the knowledge graph can represent relationships between the entities.
  • a knowledge graph G can be represented as a set of triples of the form (subject, relation, object), e.g. denoted as (s, r, o) .
  • the knowledge graph is a single united graph in which the sensor data, the sub-activities and the higher-level activities are modeled as different abstraction levels.
  • semantically matching activities may be found and connected through a two-level graph structure in order to infer an alternative type of execution for a predetermined higher-level activity associated with the collected sensor data.
  • the matching candidates may include matching sub-graphs that describe the same higher-level activity from a semantic point of view.
  • determining emission information includes an emission estimation that is treated as a missing value problem by looking at other sub-activities, preferably which occur in a similar setting, in the knowledge graph, such that overall emissions of one or more higher- level activities can be used to deduce unknown emissions.
  • the method may further comprise a step of computing which sub-activities of predetermined higher-level activities that are associated with the collected sensor data can be replaced with alternative subactivities, such that the predetermined activities’ overall goals stay the same while the emissions of said predetermined activities are reduced.
  • the term “overall goal” may refer to the higher-level activity; hence, the overall goal may be the intention or purpose why the higher-level activity is performed in the first place. By substituting sub-activities/steps of the higher-level activity, the overall goal will not change.
  • the computing of replaceable subactivities may consider user- and/or environmental-related constraints. For instance, the computation may consider requirements and/or conditions that need to be fulfilled to perform the substitute/alternative sub-activity. According to an embodiment of the invention, it may be provided that, in particular by the use of a machine learning model, a substitution for a sub-activity is learned
  • the structure can be employed for reducing emissions of a predetermined activity.
  • the structure of subactivity and higher-level activity may reflect a hierarchical concept where actions (higher-level activity) may consist of several - preferably atomic and/or intermediate
  • this hierarchical concept/structure is combined with the knowledge graph structure to learn substitutions of sub-activities by looking
  • an embodiment of the invention may encode, while learning the numerical representation in the vector space, this information in the position of the numerical representation in the vector space.
  • the distance between two numerical representations in the embedding space may reflect how similar or dissimilar two sub-activities are.
  • the angle at which they are positioned to each other may indicate the direction of the potential substitution.
  • the method may further comprise a step of providing as output a list of one or more ranked sub-activities for a predetermined sub-activity, which can be used for substitution.
  • the substitution may mean that a window is opened instead of turning on the air conditioner.
  • Embodiments of the invention overcomes the issue that people usually have a lack of knowledge or no experience in how to act carbon neutral, be it in a flat or in a company or factory.
  • Embodiments of the invention can be used in many applications where the aim is to reduce or analyze emissions. For example, it can be used in any instance where a sensor network is available that gathers information about its surrounding or the entity of interest.
  • a differentiator is that embodiments of the present invention are a short-term solution for emission reduction, i.e. , can be implemented right away. This is important because we are running out of time in the fight against (CO2) emission.
  • Fig. 1 is a schematic view illustrating an overview of a system architecture in accordance with an embodiment of the invention
  • Fig. 2 is a schematic view illustrating a Domain of Interest module according to an embodiment of the invention
  • Fig. 3 is a schematic view illustrating an Activity Isolation module according to an embodiment of the invention.
  • Fig. 4 is a schematic view illustrating an Emission Value Estimation module according to an embodiment of the invention
  • Fig. 5 is a schematic view illustrating an Input Knowledge Graph for an Action Substitution module according to an embodiment of the invention.
  • Fig. 6 is a schematic view illustrating a learning process of a machine learning model for an Action Substitution module according to an embodiment of the invention.
  • Fig. 1 shows a schematic view illustrating an overview of a system architecture in accordance with an embodiment of the present invention.
  • Embodiments of the invention may comprise several components, in particular a Domain of Interest module, an Activity Isolation module, an Activity Matching module, a module for Emission Estimation, and finally, an Action Substitution module.
  • these modules are described in more detail and how they are connected in accordance with embodiments of the invention.
  • Embodiments of the present invention have a technical aim and serve a specific technical purpose, i.e., the embodiments aim to influence non-human systems. However, for the description, it is referred for simplicity to humans as target entity.
  • Fig. 2 shows a Domain of Interest module according to an embodiment of the invention.
  • the Domain of Interest module of Fig. 2 essentially represents the data source gathered by various sensors (e.g., Inertial sensors, RFID, microphones, and cameras) which are, e.g., present in smart devices.
  • the Domain of Interest could be a person and the smart-phone, smart-watch, and smart-home provide information about the activities of that person (e.g., through an automatically generated diary).
  • the Domain of Interest provides the behavior of the entity of interest, i.e., what the entity is doing and how.
  • the collected information may be stored and forwarded (output) in a graph format.
  • the information may be represented in form of triples like ⁇ Tom> ⁇ cooks> ⁇ Dinner>, ⁇ Dinner> ⁇ consists of> ⁇ 200g Beef>, and ⁇ Beef> ⁇ was produced in> ⁇ Spain>.
  • each individual sensor signal is modeled as a triple, e.g., ⁇ sensor_event_1 > ⁇ has_value> ⁇ 10>, ⁇ sensor_event_1 > ⁇ has_unit> ⁇ m/s A 2>, ⁇ sensor_event_1 > ⁇ has_source> ⁇ accelerometer>, ⁇ sensor_event_1 > ⁇ has_timestamp> ⁇ 123456789>.
  • a set of these kind of triples, from different sensor types, which describe the same time window are fed to a set of different machine learning models to extract the higher-level meaning (in the simplest case, this can be a classification task).
  • one model is in charge to infer the performed sub-activity.
  • a subactivity is, e.g., cutting which belongs to the higher-level activity cooking.
  • Another Machine Learning (ML) model extracts the objects which caused the sensor events, or which are visible on images (captured by the camera sensor).
  • ML Machine Learning
  • the set of triples is extended by information like ⁇ sensor_event_1 > ⁇ belongs_to> ⁇ sub_activity_a>, ⁇ sensor_event_2> ⁇ describes> ⁇ object_a>, ⁇ object_a> ⁇ performs> ⁇ sub_activity_a>, etc.
  • Fig. 3 shows an Activity Isolation module according to an embodiment of the invention.
  • the Activity Isolation module takes as input the graph (i.e., set of triples) from the Domain of Interest module.
  • the Activity Isolation module has a machine learning based engine (e.g., a trained neural network) which identifies individual activities within the provided graph.
  • An individual activity e.g., ⁇ Tom> ⁇ cooks> ⁇ Dinner>
  • the machine learning module is trained on graphs to identify, i.e.
  • the approach is the same as using a link prediction algorithm for answering ⁇ Tom, has_gender, ?>, i.e., the algorithm tries to find entities which are similar to Tom, to infer the gender of Tom.
  • a link prediction algorithm for answering ⁇ Tom, has_gender, ?> i.e., the algorithm tries to find entities which are similar to Tom, to infer the gender of Tom.
  • predefined subgraphs which form an (isolated) activity.
  • the machine learning model does not simply learn to find these predefined subgraphs but rather to find those which are similar.
  • the output of Activity Isolation module may be a list of sub-graphs, each representing a particular activity, which was performed by the entity of interest.
  • an Activity Matching module takes the result of the Activity Isolation module and data of a (historical) database of activity descriptions, represented also as graphs, as input. For each sub-graph, which the Activity Matching module gets from the Activity Isolation module, it searches matching candidates in the (historical) database.
  • matching means that the matching sub-graphs describe the same activity from a semantic point of view, i.e., the actual steps (sub-activities) how the activity is performed can be different.
  • an “actual step” may refer to an atomic/intermediate step. A set of atomic/intermediate steps forms an activity, and several activities can form again another activity. There is no limitation in terms of the number of hierarchy levels. This hierarchal concept may be encoded in the knowledge graph.
  • the sub-graph provided by the Activity Isolation module might describe the activity “goes to work” where this sub-graph also covers the information that the entity of interest actually used the car.
  • a matching activity from the (historical) database would also describe the activity “goes to work” but it is possible that actually the bike was used. This means that the higher-level meaning (“goes to work”) is the same but the actual underlying (sensor) data differ (car vs. bike).
  • the Activity Isolation module may perform node label matching between nodes of type “Activity”.
  • Activity Isolation module is configured to search in the historical database for nodes that have the label “goes to work”, and are of type “activity”.
  • the output of Activity Matching module can be a set of matching candidates for each sub-graph from the Activity Isolation module.
  • Fig. 4 shows an Emission Value Estimation module (for emission estimation) according to an embodiment of the invention.
  • the Emission Value Estimation module takes the result from the Activity Matching module. In addition, it has access to the historical database, and a (preferably public) emission database. Both databases may provide (historical) data about emissions in respect of certain activities, actions or sub-activities. An integrity module may ensure, through entity resolution, that both databases do not provide contradictory information.
  • the Emission Value Estimation module analyzes each activity (provided by the Activity Matching module) and determines for each activity and sub-activity/action the emissions, based on the matching candidates and (historical) data. This may happen in two steps: First, equal and similar sub-activities are identified through link prediction (e.g.
  • Step 1 essentially infers a set of triples from the knowledge graph, which can be used to compute the missing value for a particular sub-activity.
  • this list of triples may be used to infer the emissions for sub-activity_A.
  • the (weighted) mean can be computed.
  • a machine learning (ML) method can be used.
  • the ML method treats the list of triples as features, i.e. , as input and computes or predicts the emission level for sub-activity_A. In the simplest case, this would be a classification task. The aim is to estimate or calculate for each sub-activity of an activity the emissions which in turn results in the overall emissions for a certain activity.
  • the output of the Emission Value Estimation module may be a list of activity graphs, enriched with detailed emission information (e.g., ⁇ cycling> ⁇ has_emission_level> ⁇ low>, ⁇ cycling> ⁇ has_less_emission_than> ⁇ driving>).
  • Fig. 5 shows a schematic view illustrating an Input Knowledge Graph for an Action Substitution module according to an embodiment of the invention.
  • the Action Substitution module takes the output of the emission estimation module to compute which sub-activities could be replaced by alternative sub-activities so that the overall activity stays the same but the overall emissions are reduced.
  • the sub-activity driving which could belong to the activity “going to work”, could be substituted with cycling.
  • the module may take potential (user) constraints into account.
  • requirements which need to be fulfilled to perform the substitute activity i.e. the alternative/new sub-activity
  • the potentially increased effort for the substitute activity e.g., more time required.
  • the substitute activity refers to the activity that serves as a replacement.
  • the goal of an embodiment of the invention is to find actions/activities that can substitute other actions/activities with the aim to reduce emissions and at the same time to full-fill the same task.
  • potential requirements which an activity, which can serve as a replacement may be considered.
  • the Action Substitution module may rely on the matching candidates, which were initially identified by the Activity Matching module. Further, meta information about the activities are provided by the historical database and public databases (e.g., ontologies). This includes requirements (e.g., ⁇ cycling> ⁇ has_requirement> ⁇ bicycle>) and effort (e.g., ⁇ cycling> ⁇ is_more_effort_than> ⁇ driving>).
  • the output of the Action Substitution module may be a ranked list of potential substitution (e.g., ⁇ driving> ⁇ substituted_by> ⁇ cycling>) for a particular sub-activity. All information can be modeled as triples and united into a single knowledge graph (see Fig.
  • the key functionality of the Action Substitution module is the Machine Learning model, which is tuned to learn suitable substitutions (representations) of a sub-activity so that through a link prediction task, relations like ⁇ substituted_by> can be reliable predicted.
  • the distance and the position of the embeddings in the embedding space reflect how suitable a subactivity for substituting another sub-activity is.
  • Fig. 6 illustrates the mechanism in detail.
  • Fig. 6 is a schematic view illustrating a learning process of a machine learning model for an Action Substitution module according to an embodiment of the invention.
  • substitution representation
  • two loss value are computed which measure a) the distance between the embeddings and b) the angle in the vector space. These are compared to the actual ground truth to compute the final loss value. (Personal) requirements are encoded in the provided “substitutable value” for two specific sub-activities. This value is provided along the training data.
  • the output of the Action Substitution module may be directly fed into another nonhuman system.
  • a predictive building control system or an electronic financial trading system For instance, a predictive building control system or an electronic financial trading system.
  • the following embodiments show examples:
  • Smart Building Predictive Building Control Use Case: Large companies, public buildings, or smart homes have to optimize the comfort of each thermal zone by taking into account individual thermodynamics of thermal zones and different weather impacts on the external surfaces or faces of each zone of a building. Our invention can be used to improve and determine the overall optimal building energy performance. This in turn leads to reduced (carbon) emissions.
  • Data Source A sensor network of the building which provides thermal information about the different zones and conditions of doors, windows, number of people, and available devices such as air conditioning systems.
  • a method and/or system in accordance with an embodiment of the invention helps by reaching the desired state of each thermal zone.
  • the desired state of a thermal zone can be considered as an activity that consists of several sub-activities/actions such as opening the window for five minutes, reducing the cooling level of the air conditioning for that time span, and/or lowering the heat after people left the room. While there are several possibilities to reach the desired conditions for a thermal zone, our invention selects the one which causes as few (carbon) emissions as possible.
  • Output A list of (ranked) sub-activities/actions that should be performed to reach the desired thermal zones. These can be directly feed to the predictive building control system.
  • the output of the method and/or system in accordance with an embodiment of the invention can be directly used as input to a predictive building control system, which is able to control windows, doors, heating systems, air condition systems, etc.
  • Data Source A sensor network of the farm that provides information about the weather, soil, state of the plants, and used treatment for the plants.
  • a method and/or system in accordance with an embodiment of the invention helps by finding when and how to execute certain actions when treating the plants.
  • our invention computes when would be the best point in time to do that.
  • time windows e.g., weather dependencies
  • Output A list of (ranked) sub-activities/actions which should be performed (e.g., recommendations to the farmer) to reach the desired state. These can be directly feed to the machines which have to treat the plants (e.g., a drone for watering).
  • our invention could be used to reduce the emissions of the daily routine of an individual by analyzing the daily processes through a sensor network and the smart devices.
  • the assistant can be used as a teacher for people who are willing to focus on emission reduction.
  • this activity essentially consists of sub- activities/actions like producing warm water. The best point in time for producing warm water might depend on the weather as the house has solar panels.
  • a ventilation system with heat recovery might use the heat in the bathroom when showering to manage the temperature of other rooms. For a human, it is often too difficult to capture all of these aspects and parameters to make the best decision in respect of causing as less emissions as possible.
  • Data Source Smart devices and sensor networks which capture aspects like which rooms are in use (presence sensors), which doors or windows are opened or closed (optical sensor), which items (e.g., television screen) are in use, air quality (e.g., co2 sensor), number of people, etc.
  • a method and/or system in accordance with an embodiment of the invention helps by finding when and how to execute a certain activity.
  • our invention computes when would be the best point in time so that the production of warm water and the heat recovery is most effective.
  • Output A list of (ranked) sub-activities/actions which should be performed to reduce the emissions for a certain activity. These can be directly fed to the smart home management system to automatically adapt the devices such as the ventilating system with heat recovery.
  • the output of the method and/or system in accordance with an embodiment of the invention could be directly used to adjust the level of heating, program the start time of the dish-washer or washing machine, control the charger of the e-car, control the light, control when to heat up water etc.

Abstract

L'invention concerne un procédé implémenté par ordinateur pour prendre en charge une réduction d'émissions, le procédé consistant : à collecter des données de capteur provenant de multiples capteurs d'un réseau de capteurs ; à transformer les données de capteur en un graphe de connaissances ; à déduire, à partir du graphe de connaissances, des sous-activités qui ont provoqué les données de capteur collectées, les sous-activités inférées étant ajoutées au graphe de connaissances ; à déduire, à partir du graphe de connaissances, des informations de connaissances concernant les sous-activités qui forment une activité de niveau supérieur, les informations de connaissances inférées étant ajoutées au graphe de connaissances de telle sorte que des activités de niveau supérieur isolées sont identifiées dans le graphe de connaissances ; à déterminer, dans une base de données historique, des candidats correspondants pour les activités de niveau supérieur, les candidats correspondants étant ajoutés au graphe de connaissances ; à déterminer des informations d'émissions pour chaque sous-activité et chaque activité de niveau supérieur sur la base des candidats correspondants et des données historiques, les informations d'émissions déterminées étant ajoutées au graphe de connaissances. Un système correspondant est en outre divulgué.
PCT/EP2022/067931 2022-04-05 2022-06-29 Procédé et système de prise en charge de réduction d'émissions WO2023193934A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP22166754.6 2022-04-05
EP22166754 2022-04-05

Publications (1)

Publication Number Publication Date
WO2023193934A1 true WO2023193934A1 (fr) 2023-10-12

Family

ID=81328160

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2022/067931 WO2023193934A1 (fr) 2022-04-05 2022-06-29 Procédé et système de prise en charge de réduction d'émissions

Country Status (1)

Country Link
WO (1) WO2023193934A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2858015A1 (fr) * 2013-10-04 2015-04-08 Building Research Establishment Ltd Système et procédé de simulation, commande et surveillance des performances de systèmes d'énergie
US20200082289A1 (en) * 2018-09-11 2020-03-12 Microsoft Technology Licensing, Llc Task scheduling recommendations for reduced carbon footprint
US20200372588A1 (en) * 2019-05-20 2020-11-26 Singularity Energy, Inc. Methods and systems for machine-learning for prediction of grid carbon emissions
CN113887552A (zh) * 2021-08-25 2022-01-04 苏州数言信息技术有限公司 一种基于传感器的碳计量方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2858015A1 (fr) * 2013-10-04 2015-04-08 Building Research Establishment Ltd Système et procédé de simulation, commande et surveillance des performances de systèmes d'énergie
US20200082289A1 (en) * 2018-09-11 2020-03-12 Microsoft Technology Licensing, Llc Task scheduling recommendations for reduced carbon footprint
US20200372588A1 (en) * 2019-05-20 2020-11-26 Singularity Energy, Inc. Methods and systems for machine-learning for prediction of grid carbon emissions
CN113887552A (zh) * 2021-08-25 2022-01-04 苏州数言信息技术有限公司 一种基于传感器的碳计量方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
THOMAS DIETZ ET AL.: "Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions", PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES, vol. 44, 2009, pages 18452 - 18456

Similar Documents

Publication Publication Date Title
Himeur et al. AI-big data analytics for building automation and management systems: a survey, actual challenges and future perspectives
Liang et al. Occupancy data analytics and prediction: A case study
Feng et al. Smart home: Cognitive interactive people-centric Internet of Things
Chen et al. Modeling and optimization of complex building energy systems with deep neural networks
Hagras et al. An incremental adaptive life long learning approach for type-2 fuzzy embedded agents in ambient intelligent environments
Chen et al. A novel data-driven approach for residential electricity consumption prediction based on ensemble learning
Dounis Artificial intelligence for energy conservation in buildings
CN108536030A (zh) 一种基于anfis算法的智能家居系统及其工作方法
Shah et al. Dynamic user preference parameters selection and energy consumption optimization for smart homes using deep extreme learning machine and bat algorithm
Skulimowski Freedom of choice and creativity in multicriteria decision making
Jia et al. Occupant behavior modeling for smart buildings: A critical review of data acquisition technologies and modeling methodologies
Vázquez et al. Usage profiles for sustainable buildings
CN111694280A (zh) 一种应用场景的控制系统及其控制方法
Meurer et al. Ambient intelligence for the internet of things through context-awareness
Merino et al. Metadomotic optimization using genetic algorithms
WO2023193934A1 (fr) Procédé et système de prise en charge de réduction d'émissions
May The reinforcement learning method: A feasible and sustainable control strategy for efficient occupant-centred building operation in smart cities
Daum On the Adaptation of Building Controls to the Envelope and the Occupants
Boulmaiz et al. Optimizing occupant actions to enhance his comfort while reducing energy demand in buildings
Alyafi Generation of explanations for energy management in buildings
Boulmaiz et al. A data-driven approach for guiding the occupant’s actions to achieve better comfort in buildings
Zhang Data-driven whole building energy forecasting model for data predictive control
Armero et al. Decision Making in Uncertain Rural Scenarios by means of Fuzzy TOPSIS Method.
Behl et al. Interactive analytics for smart cities infrastructures
Yashin et al. Assessment of Material and Intangible Motivation of Top Management in Regions Using Multipurpose Genetic Algorithm

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: 22743765

Country of ref document: EP

Kind code of ref document: A1