The document describes a semantic decision support system called OPTIMUS that optimizes energy use in public buildings. OPTIMUS integrates data from various sources like weather forecasts, social media, occupancy sensors, energy prices and renewable energy production. It uses a semantic framework including an ontology to model this multidisciplinary data. Prediction models use historical data to forecast building behavior. Inference rules then suggest short-term action plans based on the predictions and real-time monitored data. The system interfaces display this information to support user decisions. It was tested in pilot cities to optimize actions like HVAC scheduling and electricity self-consumption or feeding to the grid.
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Building a semantic-based decision support system to optimize the energy use in public buildings: the OPTIMUS project
1. A Semantic Decision Support System to
optimize the energy use of public buildings
Álvaro Sicilia, Gonçal Costa, Leandro Madrazo
ARC Engineering and Architecture La Salle
Ramon Llull University, Barcelona, Spain
Vincenzo Corrado, Alice Gorrino
Department of Energy
Politecnico di Torino, Torino, Italy
Fulvio Corno
Department of Control and Computer Engineering
Politecnico di Torino, Torino, Italy
2. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Outline
1. “Smart cities” paradigm for decision making
2. Decision support systems overview
3. The OPTIMUS DSS
4. The Semantic Framework
5. Conclusions / Collaboration
3. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
The “smart cities” approach can help to improve the citizens’ quality of life in
accordance with the objectives set by sustainable energy policies of the European
Union with the target of reducing by 20% the CO2 emissions by 2020
Smart cities rely on the availability of data
although there are more and more energy and other related data sets available, it
is necessary to integrate them in order to provide the various key actors
the information they need to make well-informed decisions
it is not enough to have access to the data
it is necessary to integrate data from different domains in order to
understand the interrelationships between the various areas –energy,
economics, social– that are involved in the reduction of carbon emissions in cities
1. “Smart cities” paradigm for decision making
4. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
these technologies can be applied, mainly, to support data integration
processes and to overcome the interoperability barriers between the
data generated by the different users and by the applications
There are well-known issues:
• Type of accessing
• Syntax of the content
• Data format
application of Semantic Web technologies can help to overcome some
of the difficulties which are intrinsic to the development of decision
support systems which rely on distributed and heterogeneous data
1. “Smart cities” paradigm for decision making
Data
User
Data
Data
User
User
User
User
Data Application
Systems
5. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
these technologies can be applied, mainly, to support data integration
processes and to overcome the interoperability barriers between the
data generated by the different users and by the applications
Data
User
Data
Data
User
User
User
User
Data Application
SW
Solutions for data integration with
explicit semantics can ensure that the
meaning of data can be unambiguously
understood by both humans and systems
There are well-known issues:
• Type of accessing
• Syntax of the content
• Data format
……….OWL, RDF, SPARQL
application of Semantic Web technologies can help to overcome some
of the difficulties which are intrinsic to the development of decision
support systems which rely on distributed and heterogeneous data
1. “Smart cities” paradigm for decision making
Systems
6. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Outline
1. “Smart cities” paradigm for decision making
2. Decision support systems overview
3. The OPTIMUS DSS
4. The Semantic Framework
5. Conclusions / Collaboration
7. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
2. Decision support systems overview
Classification of DSSs [1]:
- Communication-driven DSS: use a set of parameters provided by decision makers to assist them in analysing its
particular problem.
- Data-driven DSS: are based on analysing time-series data as well as external and real-time data.
- Document-driven DSS: are focused on providing search functionalities to help managers to find documents
- Knowledge-driven DSS: are based on the knowledge extraction from a particular domain to be analysed using data
mining methods,
- Model-driven DSS: operate on a model of reality rather than on data intensive model.
Use of Semantic Web Technologies for decision support:
- Semantic web technologies can be applied in DSS developments using ontologies and rules as a means to provide
intelligent support to decision-making [2].
- They can be used to support data integration processes, and to overcome the interoperability barriers through
standardized formats.
[1] Power, D. J. (2002). Decision Support Systems: Concepts and Resources for Managers. Westport, CT: Greenwood/Quorum
[2] Blomqvist, E. (2012). The Use of Semantic Web Technologies for Decision Support - A Survey, In: Semantic Web Journal, 5(3): 177-201,
IOS Press.
8. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
2. Decision support systems overview
Decision support systems in energy efficiency:
- At present, analysis techniques of energy efficiency of buildings used for decision support are limited to
a very few data sources.
- The most commonly used data sources are those provided by a building management system (e.g.,
energy consumption, temperature, humidity and CO2), while other such as weather forecasting, social
media and occupancy in most cases are not considered [3].
- However, and as a result of the increasing demand to satisfy the current legislative framework; for
example, to meet the European directives in terms of energy efficiency in buildings; new paradigms and
systems are emerging with the aim of achieving a more comprehensive view of the energy performance
of the building
- Examples of research projects: EEPOS (2015), EnRiMa (2013), SEMANCO (2013), SEMERGY
(2014), KnoholEM (2014).
[3] Corry, E., Coakley, D., O'Donnell, J., Pauwels, P. and Keane, M. (2013). The role of Linked Data and the Semantic Web in Building
Operation. Proceedings of the 13th annual International Conference for Enhanced Building Operations (ICEBO). Montréal, Canada.
9. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
EECITIES: Service platform to support planning of energy efficient cities [4]
An energy analysis service provider that supports planners, energy consultants, and policy makers to make
informed decisions related to improving the energy efficiency of urban areas.
Integration of dispersed energy related data from multiple sources, including Cadastre, census, socio-
economic, building typologies (u-values, windows properties, systems)
[4] Madrazo, L., Nemirovski, G., Sicilia. A. (2013). Shared Vocabularies to Support the Creation of
Energy Urban Systems Models. In Proceedings 4th Workshop organised by the EEB Data Models
Community ICT for Sustainable Places, Nice, France, September, 2013.
from: http://www.eecities.com
OWL 2 QL profile
SUMO Ontology, R2RML mappings,
Query federation,
http://www.eecities.com/
Services:
- Assessing the current energy performance of
buildings in towns and cities.
- Identifying priority areas and buildings for
energy efficiency interventions.
- Evaluating the impact of refurbishing
buildings at the urban level.
- …
2. Decision support systems overview
10. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Outline
1. “Smart cities” paradigm for decision making
2. Decision support systems overview
3. The OPTIMUS DSS
4. The Semantic Framework
5. Conclusions / Collaboration
11. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Semantic framework
Weather
forecasting
De-centralized
sensor-based
Social media/
mining
Energy
prices
RES
production
Data is captured from the buildings and their
context. Semantic framework integrates the
different data sources using semantic web
technologies.
DATA CAPTURING MODULES
3. The OPTIMUS DSS
12. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Semantic framework
Weather
forecastin
g
De-centralized
sensor-based
Social media/
mining
Energy
prices
RES
production
PREDICTION
MODELS
Historical data
Predicted data
Prediction models use historical data
to forecast the building behaviour
for the following 7 days
DATA CAPTURING MODULES
3. The OPTIMUS DSS
13. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Semantic framework
Weather
forecastin
g
De-centralized
sensor-based
Social media/
mining
Energy
prices
RES
production
PREDICTION
MODELS
INFERENCE RULES
Historical data
Predicted data
Inference rules use the predicted
and monitored data to suggest
short-term actions to the final user
Monitored data
DATA CAPTURING MODULES
Relations between input data (real time and predicted data,
and static user inputs) for suggesting an action plan
3. The OPTIMUS DSS
14. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Semantic framework
Weather
forecastin
g
De-centralized
sensor-based
Social media/
mining
Energy
prices
RES
production
PREDICTION
MODELS
INFERENCE RULES
Historical data
Predicted data
Monitored data
DATA CAPTURING MODULES
DSS interfaces display the
monitored data, forecasted data,
and short-term plans in order to
support users’ decisions
DSS INTERFACE
Relations between input data (real time and predicted data,
and static user inputs) for suggesting an action plan
3. The OPTIMUS DSS
15. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Semantic framework
Weather
forecastin
g
De-centralized
sensor-based
Social media/
mining
Energy
prices
RES
production
PREDICTION
MODELS
INFERENCE RULES
Historical data
Predicted data
Monitored data
DATA CAPTURING MODULES
Sant Cugat
Savona
Zaanstad
The results of the implementation
of the actions in each pilot city will
modify the data sources
DEMONSTRATIONDSS INTERFACE
Relations between input data (real time and predicted data,
and static user inputs) for suggesting an action plan
3. The OPTIMUS DSS
16. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
The OPTIMUS DSS optimizes the use of energy, suggesting ACTION PLANS:
• boost time of the heating/cooling system taking into account the forecasting of the
outdoor/indoor air temperature and the occupancy of the building.
• selling/self-consuming of electricity produced by a PV system considering different
scenarios of energy market and strategies (green, finance, peak).
• adjustment of the temperature set-point, taking into consideration thermal comfort
parameters (e.g., Predicted Mean Vote index) using occupants’ inputs gathered with a
mobile app.
3. The OPTIMUS DSS
17. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
The innovative approach of the OPTIMUS DSS is based on the combination of:
• The use of multidisciplinary data sources, including:
• Weather forecasting
• Social media
• Occupancy
• The semantic modelling of data using Semantic Web technologies
• The integration of data for energy optimization
3. The OPTIMUS DSS
18. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Outline
1. “Smart cities” paradigm for decision making
2. Decision support systems overview
3. The OPTIMUS DSS
4. The Semantic Framework
5. Conclusions / Collaboration
19. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Social media
Sources
Data capturing
modules
Virtuoso Server
Weather data
De-centralized
data
Energy prices
Energy
production data
Semantic
Framework
RapidAnalytics:
Prediction models
Semantic Service
Ztreamy Server
PHP Services:
Inference rules
MariaDB
Data portal. Elda: the
linked-data API in Java
End-user web
environment
Management
environment
DSS engine
Developed within OPTIMUS project External source used by OPTIMUS
Sources Sources Sources Sources
DSS environments
R scripts
Internal architecture
of the OPTIMUS DSS
20. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Social media
Sources
WP2 Data capturing
modules
Virtuoso Server
Weather data
De-centralized
data
Energy prices
Energy
production data
RapidAnalytics:
Prediction models
Semantic Service
Ztreamy Server
PHP Services:
Inference rules
MariaDB
Data portal. Elda: the
linked-data API in Java
End-user web
environment
Management
environment
DSS engine
Developed within OPTIMUS project External source used by OPTIMUS
Sources Sources Sources Sources
DSS environments
R scripts
4. The Semantic Framework
Semantic
Framework
21. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
We have devised a semantic framework which is composed of:
1. A shared conceptualization of the urban and building domain including
monitoring devices, formally implemented as the OPTIMUS ontology coded in
OWL.
2. A semantic integration process for capturing and modelling data sources from
different domains.
- Two RDF templates used by the data capturing modules for modelling real-
time information items, according to the OPTIMUS ontology.
- A publish-and-subscribe system as a communication infrastructure between
the data capturing modules and the DSS implemented with the Ztreamy system
and a semantic service which processes the data with the purpose of
contextualizing them.
22. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Sensors
(based on SSN ontology)
OPTIMUS ontology
Building & systems features
(based on Semanco ontology)
1. The OPTIMUS ontology reuse two existing ontologies:
- Static data (Building and systems features) has been modelled with the
SEMANCO ontology
http://semanco-tools.eu/ontology-releases/eu/semanco/ontology/SEMANCO/SEMANCO.owl
- Dynamic data (sensoring) has been modelled with the
Semantic Sensor Network (SSN) ontology
http://purl.oclc.org/NET/ssnx/ssn
23. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Ontologies
ssn:Sensor
ssn:SensingDevice
ssn:Observation
optimus:SunnyPortal_EnergyProduction
semanco:Solar_Irradiationssn:FeatureOfInterest
ssn:Property
ssn:System
subClassOf
subClassOf
ssn:hasSubSystem
ssn:observes
ssn:observes
subClassOf
ssn:hasProperty
subClassOf
ssn:observedBy
subClassOf
ssn:featureOfInterest
ssn:observedProperty
semanco:PVSystem_Peak_Power
optimus:SunnyPortal_SolarRadiation
subClassOf
ssn:SensorOutput
ssn:observationResult
ssn:hasValue
time:Instant
ssn:observationResultTime
time:inXSDDateTime
literal
ssn:Platform
ssn:Deployment
ssn:deployedOnPlatform
ssn:hasDeployment
sumo:located
sumo:Building
sumo:Room
Semanco:Space_Heating_System
Semanco:Ventilation_System
…
literal
subClassOf
optimus:Solar_IrradiationSensorOutput
optimus:PVSystem_Peak_PowetSensorOutput
subClassOf
ssn:onPlatform
optimus:SunnyPortal
subClassOf
subClassOf
ssn:observes
optimus:Solar_IrradiationFeature
subClassOf
optimus:PVSystem_Peak_PowerFeature
ssn:hasProperty
Static part of the ontology
Building and System features
Semantic Sensor Network
OPTIMUS
SEMANCO
24. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Triple store
(Virtuoso Server)
Semantic Framework
Weather data
De-centralized
data
Social media
Ztreamy
Server
Energy prices
Energy
production
Semantic
Service
Data capturing modulesSources
OPTIMUS DSS
2. a semantic integration process for capturing and modelling data sources
from different domains.
RAW
DATA
RDF DATA:
RAW DATA + MEANING
RDF DATA +
CONTEXT
INTEGRATED
DATA
25. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Triple store
(Virtuoso Server)
Semantic Framework
Weather data
De-centralized
data
Social media
Ztreamy
Server
Energy prices
Energy
production
Semantic
Service
Data capturing modulesSources
OPTIMUS DSS
1. Data
translation
RAW
DATA
26. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Triple store
(Virtuoso Server)
Semantic Framework
Weather data
De-centralized
data
Social media
Ztreamy
Server
Energy prices
Energy
production
Semantic
Service
Data capturing modulesSources
OPTIMUS DSS
1. Data
translation
2. Data
communication
RAW
DATA
RDF DATA:
RAW DATA + MEANING
publishers
27. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Triple store
(Virtuoso Server)
Semantic Framework
Weather data
De-centralized
data
Social media
Ztreamy
Server
Energy prices
Energy
production
Semantic
Service
Data capturing modulesSources
OPTIMUS DSS
1. Data
translation
2. Data
communication
RAW
DATA
RDF DATA:
RAW DATA + MEANING
RDF DATA +
CONTEXT
publishers
Subscriber
3. Data
contextualization
28. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Triple store
(Virtuoso Server)
Semantic Framework
Weather data
De-centralized
data
Social media
Ztreamy
Server
Energy prices
Energy
production
Semantic
Service
RAW
DATA
RDF DATA:
RAW DATA + MEANING
RDF DATA +
CONTEXT
INTEGRATED
DATA
Data capturing modulesSources
OPTIMUS DSS
1. Data
translation
2. Data
communication
3. Data
contextualization
4. Data
storage
publishers
Subscriber
29. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Ontologies
ssn:Sensor
ssn:SensingDevice
ssn:Observation
optimus:SunnyPortal_EnergyProduction
semanco:Solar_Irradiationssn:FeatureOfInterest
ssn:Property
ssn:System
subClassOf
subClassOf
ssn:hasSubSystem
ssn:observes
ssn:observes
subClassOf
ssn:hasProperty
subClassOf
ssn:observedBy
subClassOf
ssn:featureOfInterest
ssn:observedProperty
semanco:PVSystem_Peak_Power
optimus:SunnyPortal_SolarRadiation
subClassOf
ssn:SensorOutput
ssn:observationResult
ssn:hasValue
time:Instant
ssn:observationResultTime
time:inXSDDateTime
literal
ssn:Platform
ssn:Deployment
ssn:deployedOnPlatform
ssn:hasDeployment
sumo:located
sumo:Building
sumo:Room
Semanco:Space_Heating_System
Semanco:Ventilation_System
…
literal
subClassOf
optimus:Solar_IrradiationSensorOutput
optimus:PVSystem_Peak_PowetSensorOutput
subClassOf
ssn:onPlatform
optimus:SunnyPortal
subClassOf
subClassOf
ssn:observes
optimus:Solar_IrradiationFeature
subClassOf
optimus:PVSystem_Peak_PowerFeature
ssn:hasProperty
Static part of the ontology
Building and System features
Semantic Sensor Network
OPTIMUS
SEMANCO
RAW
DATA
30. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Ontologies
ssn:Sensor
ssn:SensingDevice
ssn:Observation
optimus:SunnyPortal_EnergyProduction
semanco:Solar_Irradiationssn:FeatureOfInterest
ssn:Property
ssn:System
subClassOf
subClassOf
ssn:hasSubSystem
ssn:observes
ssn:observes
subClassOf
ssn:hasProperty
subClassOf
ssn:observedBy
subClassOf
ssn:featureOfInterest
ssn:observedProperty
semanco:PVSystem_Peak_Power
optimus:SunnyPortal_SolarRadiation
subClassOf
ssn:SensorOutput
ssn:observationResult
ssn:hasValue
time:Instant
ssn:observationResultTime
time:inXSDDateTime
literal
ssn:Platform
ssn:Deployment
ssn:deployedOnPlatform
ssn:hasDeployment
sumo:located
sumo:Building
sumo:Room
Semanco:Space_Heating_System
Semanco:Ventilation_System
…
literal
subClassOf
optimus:Solar_IrradiationSensorOutput
optimus:PVSystem_Peak_PowetSensorOutput
subClassOf
ssn:onPlatform
optimus:SunnyPortal
subClassOf
subClassOf
ssn:observes
optimus:Solar_IrradiationFeature
subClassOf
optimus:PVSystem_Peak_PowerFeature
ssn:hasProperty
Static part of the ontology
Building and System features
Semantic Sensor Network
OPTIMUS
SEMANCO
RDF template for data capturing modules
RDF DATA:
RDF DATA + MEANING
RAW DATA
31. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
Ontologies
ssn:Sensor
ssn:SensingDevice
ssn:Observation
optimus:SunnyPortal_EnergyProduction
semanco:Solar_Irradiationssn:FeatureOfInterest
ssn:Property
ssn:System
subClassOf
subClassOf
ssn:hasSubSystem
ssn:observes
ssn:observes
subClassOf
ssn:hasProperty
subClassOf
ssn:observedBy
subClassOf
ssn:featureOfInterest
ssn:observedProperty
semanco:PVSystem_Peak_Power
optimus:SunnyPortal_SolarRadiation
subClassOf
ssn:SensorOutput
ssn:observationResult
ssn:hasValue
time:Instant
ssn:observationResultTime
time:inXSDDateTime
literal
ssn:Platform
ssn:Deployment
ssn:deployedOnPlatform
ssn:hasDeployment
sumo:located
sumo:Building
sumo:Room
Semanco:Space_Heating_System
Semanco:Ventilation_System
…
literal
subClassOf
optimus:Solar_IrradiationSensorOutput
optimus:PVSystem_Peak_PowetSensorOutput
subClassOf
ssn:onPlatform
optimus:SunnyPortal
subClassOf
subClassOf
ssn:observes
optimus:Solar_IrradiationFeature
subClassOf
optimus:PVSystem_Peak_PowerFeature
ssn:hasProperty
Static part of the ontology
Building and System features
Semantic Sensor Network
OPTIMUS
SEMANCO
RDF data from modules + context data
RAW DATA
RDF DATA:
RDF DATA + MEANING
RDF DATA +
CONTEXT
32. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
33. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
34. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
4. The Semantic Framework
35. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Outline
1. “Smart cities” paradigm for decision making
2. Decision support systems overview
3. The OPTIMUS DSS
4. The Semantic Framework
5. Conclusions / Collaboration
36. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
5. Conclusions
The OPTIMUS DSS represents an innovation with respect to the existing systems in so
far it is able to interlink five types of heterogeneous and dynamic data sources in
order to suggest short-term actions plans that enable public authorities to reduce
energy consumption in public buildings
In the semantic framework proposed in OPTIMUS, the use of RDF templates have
been an important mechanism to provide a standardized way of integrating
heterogeneous data sources. However, a monitoring process is required from the
beginning to ensure that developers of capture modules are following the specification
of the template.
Requirements capturing is not a one-time process. It needs to be continuously validated
with end-users and domain experts
37. Gonçal Costa – ARC, La Salle
October 27th -29th, 2015 Eindhoven, The Netherlands
Gonçal Costa
ARC Engineering & Architecture
La Salle, Ramon Llull university
Quatre Camins, 2 08022,
Barcelona, SPAIN
Tel. +34 93 290 24 49
Fax +34 93 290 24 20
E-mail: gcosta@salleurl.edu
Editor's Notes
OPTIMUS project is in the second year of development
Tres action plans funcionando