An Ontology for Solar Irradiation Forecast Models

Abhilash Kantamneni

and Laura E. Brown

Department of Geography, University of Guelph, 50 Stone Road E., Guelph, Canada

Department of Computer Science, Michigan Technological University, 1400 Townsend Drive, Houghton, MI, U.S.A.

Keywords:

Ontology, Solar Irradiation, Forecasting, Renewable Energy, Microgrids.

Abstract:

The growth of solar energy resources in recent years has led to increased calls for accurate forecasts of solar

irradiance for the reliable and sustainable integration of solar into the national grid. A growing body of aca-

demic research has developed models for forecasting solar irradiance, identiﬁed metrics for comparing solar

forecasts, and described applications and end users of solar forecasts. Ontologies are explicit and formal voca-

bulary of terms and their relationships that facilitate better communication, improve interoperability, and reﬁne

knowledge reuse by experts and users of the domain. This paper describes a step towards using ontologies to

describe the knowledge, concepts, and relationships in the domain of solar irradiance forecasting to develop a

shared understanding for diverse stakeholders that interact with the domain. A preliminary ontology on solar

irradiance forecasting, SF-ONT, was created and validated on three use cases.

1 INTRODUCTION

Spurred by declining photovoltaic (PV) module pri-

ces, favorable government policies, and growing con-

cerns about mitigating climate change, recent years

have seen a rapid growth in the proliferation of solar

electric systems. As PV markets continue to grow, the

variable, intermittent and non-dispatchable nature of

solar energy introduces additional uncertainty and va-

riability in grid operations (Widiss and Porter, 2014).

High-precision forecasts of solar energy output can

prove crucial for the reliable, affordable and sustaina-

ble grid integration of solar electric systems (Diagne

et al., 2013).

Solar irradiance forecasting (a proxy for solar

energy forecasting) is an emerging knowledge dom-

ain that integrates expertise from diverse ﬁelds - rese-

archers advancing forecast models, regulatory agen-

cies developing performance characteristics, deve-

lopers and practitioners integrating solar into the

grid and so on. Due to differing technical back-

grounds, expertise, knowledge hierarchies, termino-

logies, technical knowledge, and expectations, these

diverse stakeholders may lack a shared understanding

of the domain in which they interact.

Modern ontologies have emerged as a way to

share common understanding of structure of infor-

mation between communities of interest, either hu-

man or software agents. By separating domain kno-

wledge from operational, ontologies promote intero-

perability, translating between different methods, mo-

dels and paradigms (Noy et al., 2001).

A literature review reveals no comprehensive se-

mantic ontology or application of semantic ontolo-

gies to represent information and knowledge about

the domain of solar irradiance forecasting. In anti-

cipation of a rapidly growing market in solar energy

generation and integration in the electric grid, we pre-

sent a solar forecasting ontology (SF-ONT) that co-

vers solar forecasting models, their performance me-

trics, likely end-users, and grid applications. This on-

tology is then is validated on hypothetical use cases

likely to be experienced by sample end-users.

2 RELATED WORK

To develop an ontology for solar forecasting it is ne-

cessary to understand about the forecasting problem

itself, the many models that have been applied to this

problem, the users and entities that may use the mo-

dels, applications for which the models help answer

questions, metrics to compare models, and what other

ontologies exist for these topics. Each of these points

will be brieﬂy addressed in this section.

Solar energy output of a solar PV system is a

function of physical characteristics of the solar PV

system and the short-wave solar radiation ﬂux inci-

dent on it’s surface (McEvoy et al., 2011). This solar

Kantamneni, A. and Brown, L.

An Ontology for Solar Irradiation Forecast Models.

DOI: 10.5220/0006937202630270

In Proceedings of the 10th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2018) - Volume 2: KEOD, pages 263-270

ISBN: 978-989-758-330-8

263

irradiance is a sum of many parts. At the top of the at-

mosphere, solar radiation is constant over time. Cloud

cover, aerosol and dust particles absorb and scatter ra-

diation as solar irradiance passes through the atmos-

phere. Consequentially, the solar irradiance that rea-

ches any point on earth’s surface is a function of at-

mospheric and weather conditions above the location

of interest. All solar irradiance forecast models offer

a means to capture this relationship.

We review widely used forecast models, with con-

sideration given to their temporal and spatial forecast

horizons (Figure 1). Persistence Models are naive

models that assume solar irradiance at current time-

step persists till the next; used for short term fore-

casts to benchmark other models. Empirical Models

are based on empirical observations, not the physi-

cal relationships among inputs to the model; exam-

ples include sunshine based, temperature based, and

ASHRAE models (Paulescu, 2008).Radiative Models

employ remote sensing instruments on satellites or

ground measuring stations that model irradiance as

a function of altitude, location and atmospheric con-

ditions on clear sunny days (Ruiz-Arias and Guey-

mard, 2018). Time Series Models forecast trends in

solar irradiance based on statistical modeling of ob-

served patterns in the past. Several categories of mo-

dels include ARMA, ARIMA, ARIMAX, seasonal-

ARIMA, etc. (Prema and Rao, 2015).

Artiﬁcial Neural Network (ANN) Models are gene-

ral purpose computational intelligence machines that

can be trained to learn and recognize patterns using

atmospheric variables and time-series irradiance data

as inputs (Voyant et al., 2017). Cloud Imagery Mo-

dels use satellite derived models or ground-based sky

imaging cameras to estimate cloud cover as an inverse

proxy for solar irradiance (Barbieri et al., 2017). Nu-

merical Weather Prediction (NWP) Models treat the

atmosphere as a ﬂuid, using thermodynamics to esti-

mate the state of the ﬂuid in the near future and pro-

ducing forecasts of about 125 weather variables inclu-

ding solar irradiance (Reikard et al., 2017).

Accurate solar forecasting has applications offe-

ring value to multiple stakeholders in the electric

grid. Long-term forecasts of utility-scale solar may

be used for reliability planning and scheduling gene-

ration sources. Medium-term forecasts of roof-top so-

lar at the distribution end may be employed in fore-

casting demand within a load-serving entity’s service

territory. Competitive electric markets may use short-

term solar forecasts for bidding and trading energy

services. Table 1 offers a summary of solar forecas-

ting applications, end users and temporal domains in

the context of the North American electric grid.

Questions stemming from a speciﬁc grid opera-

Figure 1: Spatial and temporal domains of solar irradiance

models, adapted partly from (Pelland et al., 2013; Diagne

et al., 2013).

tions application can be linked to a irradiance fore-

casting model type via the temporal domain. For

example, a utility performing systems planning out

for several years (long term horizon) would not want

to consider a persistence irradiance forecast model

(which is valid for short term time scales). Similar

links via the spatial domain are also applicable.

Accurate solar forecasts facilitates more efﬁcient

integration of solar into utility resources. It is im-

portant to develop metrics to measure and assess the

impact of forecasts have on integrating solar into the

national grid (Zhang et al., 2015). The United Sta-

tes’ Department of Energy Sunshot Initiative identi-

ﬁes criteria that makes a solar forecasting model use-

ful - simple and easily understandable; provides acti-

onable insights; input data is manageable and acqui-

rable; and practical for operational and planning deci-

sions.

The solar forecasting ontology presented here is

novel to the literature, however, a few other related

ontologies exist (note, ontologies which are incorpo-

rated into SF-ONT are discussed in detail in Sec. 4.1).

Solar radiance is featured as one of many environ-

mental or meteorological variables in ontologies like

EEPSA (Esnaola-Gonzalez and Bermudez, 2015)

and SREQP (S

anchez-Cervantes et al., 2016), wit-

hout description of crucial attributes like forecast mo-

dels, duration, resolution and so on. While SF-ONT

describes models for solar energy generation, ontolo-

gies like Fiesta-Ont (FIESTA-IoT, 2017) and Power-

ont (Bonino et al., 2015) describe smart-home appli-

ances and other demand side energy management ser-

vices. As smart-grid research and development con-

tinues to drive support for uniﬁed semantic ontology

for energy management (Cuenca et al., 2017), future

updates to SF-ONT and related ontologies may im-

KEOD 2018 - 10th International Conference on Knowledge Engineering and Ontology Development

264

Table 1: Solar forecasting end-users, applications, and time-horizon; adapted from (Widiss and Porter, 2014; Zieher et al.,

2015).

Users Applications Time-Horizon

ISO/RTO

Reliability planning Long Term

Congestion management Medium Term

Unit commitment & dispatch Short & Medium Term

Load-ﬂow, ramps & curtailment Short & Medium Term

Security. maintenance & outage Medium & Long Term

Distribution Utilities

System planning Long Term

Outage management Medium Term

Load forecasting Medium Term

Smart Grid management Short & Medium Term

Load Serving Entity Scheduling & balancing Short & Medium Term

Energy Traders Bidding strategies Short & Medium Term

Research labs

Integration & simulation studies All Terms

Project developers

prove their interconnection and interoperability.

Some recent work describes the role of forecasts

in managing power ﬂow in smart grids. For instance,

authors in (Maffei et al., 2018) describe an Artiﬁ-

cial Neural Network (ANN) model that uses hourly

energy and weather data to forecast solar irradiance.

However, they do not offer a publicly available on-

tology, or provide insights into how this domain was

mapped into the semantic vocabulary. Furthermore,

the particular forecast model described in this paper

with its speciﬁc inputs, outputs and other attributes

can be easily modeled as an instance of the super-

class ANNmodels in SF-ONT.

3 ONTOLOGIES

The domain of solar forecasting integrates know-

ledge from ﬁelds like cloud physics, statistical me-

chanics, artiﬁcial intelligence, grid-operations, and

energy markets. Due to differing technical back-

grounds, expertise, knowledge hierarchies, agents in

the same environment may lack a shared understan-

ding of the domain in which they interact. Natural

differences in the deﬁnitions of concepts, structures,

objects and relationships may lead to poor communi-

cation, reduced interoperability, and system integra-

tion challenges (Uschold and Gruninger, 1996). On-

tologies have emerged as means to preserve interope-

rability, formally modeling domain knowledge using

instances, sets of concepts, and relationships between

sets of concepts (Guarino et al., 2009). While the fe-

atures of an ontology can also vary by the language

used to describe the ontology, this paper uses Web

Ontology Language (OWL) to represent domain kno-

wledge of solar forecasting models.

Web Ontology Language (OWL) (McGuinness

et al., 2004) is a formal semantic language from the

World Wide Web Consortium (W3C) that offers a rich

set of differential logic operators (e.g., intersection,

union), allowing for complex concepts and relations-

hips to be built from simpler concepts. Any domain

can be modeled through a shared vocabulary of indi-

viduals, properties and classes. Individuals represent

objects in the domain of interest, explicitly deﬁned as

same or different to each other. Properties are the fe-

atures and attributes of Individuals. Classes deﬁned

by the relationships between Individuals, or through

a restriction on Properties.

3.1 Ontological Development Methods

Many approaches have been proposed for formally

developing ontologies. Uschold and King is among

the ﬁrst ontology development methods publis-

hed (Uschold et al., 1998). The approach uses an ite-

ration through the following stages - identify domain

vocabulary, purpose, intended use, end-users, scope,

terms; capture and code textual concepts and relati-

onships into formal ontological language; evaluate re-

quirements; and document results from each stage to

aid the next iteration of ontology development.

SENSUS is a natural language based ontology de-

veloped to provide a broad conceptual structure for

work in machine translation. Rather than develop a

step by step or iterative process, SENSUS outlines

general principles for designing an ontology - Do not

over-commit on representational choices; extend ba-

sed on actual use conditions; integrate horizontally

with other ontologies; and cluster concepts to struc-

ture ontologies (Swartout et al., 1996).

METHODONTOLOGY acquires, evaluates and

documents domain knowledge through an iterative

process of ontology speciﬁcation (vocabulary, pur-

pose, scope, term, and other factors); conceptuali-

zation (concepts, verbs, class attributes and instances

of classes); integration (reusing existing ontologies);

and implementation (check for incomplete, inconsis-

An Ontology for Solar Irradiation Forecast Models

265

tent and redundant knowledge) (Gruber, 1993)

This paper uses Ontology 101 (Noy et al., 2001)

and the Prot ´eg ´e software tool (Musen, 2015) to de-

velop an ontology for solar forecasting. This method

was chosen for its accessibility to modelers with limi-

ted prior experience and its ﬂexibility of informal gui-

delines, providing multiple entry points into the pro-

cess of domain knowledge translation.

4 SOLAR FORECASTING

ONTOLOGY

The solar forecasting ontology (SF-ONT) developed

in this paper integrates knowledge from academic pa-

pers, grey literature, as well as industry, government,

and expert papers. This ontology is intended to be

used by project developers to compare solar forecas-

ting models, and choose an appropriate model for

their use-case scenario, while working within the li-

mitations of data and instrumentation availability.

4.1 Domain Knowledge Description

We will walk through a brief overview of the creation

of the SF-ONT.

Competency Questions: Table 2 enumerates a

sample of competency questions for illustrative pur-

poses and identiﬁes the important terms that need to

be explained to the user. Our ontology will make ’sta-

tements’ about these terms through relationships be-

tween them. At this stage, terms and relationships

are expressed in the natural language without wor-

rying about overlap of concepts. Competency questi-

ons were developed with collaborators for a research

project on microgrids.

Related Ontologies: Reusing existing and valida-

ted ontology saves time and effort. Many concepts

deﬁned in an different ontology can be directly impor-

ted, with little modiﬁcation, and applied to the dom-

ain of interest. The following formal ontologies were

considered for reuse in SF-ONT.

• OWL-Time (W3, 2017) is a list of temporal con-

cepts built for describing time related content of

Web pages like Intervals - spans of time that have

an beginning and an end, and Instants - intervals

with zero length that are used to express notions

of time time like duration, begins, ends, etc.

• OWL-Basic-GEO (W3, 2003) is a Resource Des-

cription Foundation (RDF) vocabulary that provi-

des semantic deﬁnitions for basic geographic con-

cepts like latitude, longitude and other spatial con-

cepts.

• Units of Measurement (Rijgersberg et al., 2013)

is an OWL ontology of the domain of measures

expressed in terms of a base set of SI system of

units. For example, electromagnetic irradiance is

W M

−2

• Weather ontologies that describe concepts from

the domains of weather and weather forecas-

ting were considered (W3, 2005; Gajderowicz,

2008; Staroch, 2013). However, these ontologies

describe weather variables like sunrise time and

windspeeds that are not relevant for SF-ONT’s

purposes. Since these ontologies reused modiﬁed

concepts of OWL-Time and Units, we sought to

avoid the risk of obfuscating domain knowledge

due to overlapping classes, predicates and con-

cepts.

• Concentrated Solar Power ontology, ﬁrst deve-

loped by (Piazza and Faso, 2014), largely serves

as a proof of concept for formally representing

knowledge of solar radiation modeling and fore-

casting by the means of ontologies. However, the

domain of this ontology is limited to just one type

of solar energy systems - concentrated solar power

systems, and does not include expertise on end-

users and grid applications. Nevertheless, con-

cepts like temporal and spatial domain of forecast

models were reused for SF-ONT.

Deﬁning Classes and Hierarchy: Using compe-

tency questions motivated above (examples in Ta-

ble 2), we enumerated a list of concepts (terms and

their properties), described either in terms of clas-

ses or individual instances. Classes are organized

into a hierarchical taxonomy, where subclasses in-

herit the properties of their superclass and instance

of a subclass by deﬁnition will be an instance of the

superclass; see example in Figure 2. ANN is a sub-

class that inherits the properties and relationships of

the class ForecastModels - hasInputs & hasOutputs.

ANN

represents an individual instance of a class of

ANN models such as one implemented in (Cuenca

et al., 2017).

Deﬁning Properties and Relationships: Classes

can also be deﬁned by the relationships between con-

cepts. In Figure 1, short, medium and long terms are

the temporal domains of forecast models can repre-

sented by classes and relations as shown in Figure 3.

Relationships are explicitly encoded in the develop-

ment process of the ontology where models like NWP

are subclasses of the class ForecastModels, but are

connected to the class MediumTerm through the rela-

tion hasTemporalDomain. Overall, the summary sta-

tistics and top-level classes of SF-ONT are presented

in Figure 4. Full details of SF-ONT are available at

http://pages.mtu.edu/ lebrown/research/sf-ont/.

KEOD 2018 - 10th International Conference on Knowledge Engineering and Ontology Development

266

Table 2: A sample of competency questions.

Competency questions Class Property Class

What data do I need for my model to work? ForecastModels hasInputs ForecastInputs

What models can I use with these weather variables? ForecastModels hasOutputs ForecastOutputs

Can I use this model at this location? ForecastModels hasGeographicAttr Lat & Long.

Who will use my forecast model and for what? EndUsers isResponsibleFor Applications

Figure 2: Example of hierarchy of classes in the ontology.

Figure 3: Example of relation between classes.

4.2 Testing and Validation of SF-ONT

Once domain knowledge has been explicitly speci-

ﬁed, the semantics of an ontology are veriﬁed to avoid

overlap of concepts and relationships. Simple reaso-

ners check for logical consistency by testing if ob-

jects and properties are linked correctly based on de-

ﬁned rules and axioms. In the hierarchical model of

classiﬁcation, the Prot ´eg ´e software environment that

supports asserted hierarchies that are manually named

and explicitly constructed; or inferred hierarchies that

are that are not explicitly stated in the data model but

instead are inferred using advanced reasoners. Such

advanced reasoners help populate relationships bet-

ween individuals in a domain without explicitly enco-

ding every single relationship from the domain.

Beyond such logical checks, a systematic evalu-

ation can help users make informed decisions about

choosing an ontology that best ﬁts their needs. Onto-

Figure 4: SF-ONT summary statistics and classes expres-

sing top-level concepts in solar irradiance forecasting.

logies should be understandable, and offer a concep-

tual foundation for a range of anticipated uses. On-

tologies have to be further validated to test that they

address the requirements that motivated their crea-

tion. Validation of ontologies through illustrating use-

cases is a common practice to determine if an onto-

logy is accurate, adaptable and clear (Staab and Stu-

der, 2013). Accurate ontologies comply to the know-

ledge experts of the domain, and correctly represent

the concepts of the world. The use-cases that fol-

low were developed in consultation with collabora-

tive research projects on microgrids and by no means

exhaustive, but serve as a means to illustrate the qua-

lity of the ontology.

4.2.1 Identifying Appropriate End-users based

on Constraint on Forecast Models

Consider a real world scenario where a research lab

develops an ANN based solar forecasting model, and

is interested in identifying the end-users that would

beneﬁt from such a model. In our domain knowledge

of solar forecasting, end-users are not explicitly rela-

ted to forecast models. However, in the Prot ´eg ´e soft-

ware environment, expert reasoners like FACT++ can

be used to infer a relationship. For this use case, a new

subclass ANNEndUsers, a subclass of EndUsers is

speciﬁed using the class hierarchy relationships show

in Figure 5. Figure 6 illustrates the populated inferred

dummy class.

An Ontology for Solar Irradiation Forecast Models

267

Figure 5: The asserted class relationships of a dummy class

called ANNEndUsers.

Figure 6: The inferred hierarchy of the ANNEndUsers

class.

4.2.2 Identifying Appropriate Grid Applications

based on Data Constraints

Consider a scenario where a solar developer wants to

develop a forecast model to inform speciﬁc grid ap-

plication, but is constrained by the kind of data varia-

bles available as forecast model inputs. In our domain

knowledge of solar forecasting, grid applications are

not explicitly related to input variables of forecast mo-

dels. However, expert reasoners can be used to infer

a relationship. For this case, a dummy class of appli-

cations that may use solar irradiance forecasts when

only parametric constants are available is created (Fi-

gure 7) with the inferred hierarchy shown in Figure 8.

4.2.3 Selecting Appropriate Models based on

Constraints on End-users

Consider a real world scenario where a programmer

wants to develop solar forecasting tools for a speciﬁc

end user, a Load Serving Entity (LSE) for instance.

In this use case scenario, the developer is interested

in identifying all solar forecasting models that could

be used to forecast solar irradiance according to pa-

rameters that could most beneﬁt their client. At this

stage of the development process, the developer does

not have the domain expertise to identify appropri-

ate grid applications for their forecast model. A new

subclass ModelsForLSE is created under Forecast-

Models to represent the desired class (see Figure 9).

The inferred hierarchy of ModelsForLSE is shown in

Figure 10.

Figure 7: Example of identifying grid application based on

data constraints - ParametricConstraintsApplications.

Figure 8: The inferred hierarchy of the ParametricCon-

straintsApplications class.

5 CONCLUSION AND FUTURE

WORK

This paper describes the basics of solar irradiance

and forecasting, reviews academic literature on recent

advancements in improved solar forecasting models,

and identiﬁes grid applications and end-users that will

beneﬁt from accurate solar forecasts. In doing so,

the paper integrates knowledge from a growing body

of academic research advancing models for forecas-

ting solar irradiance, with expertise from practitioners

developing applications for integrating solar into the

grid.

Due to differing technical backgrounds, expertise,

knowledge hierarchies, terminologies, technical kno-

wledge, and expectations, the diverse stakeholders

in the world of solar forecasting may lack a shared

understanding of the domain in which they interact.

In anticipation of a rapidly growing market in solar

energy generation and integration, this paper descri-

bes a step towards improving communication, intero-

perability and knowledge-sharing about solar forecas-

ting, using ontologies.

Subsequently, the paper describes ontologies and

brieﬂy reviews methodologies for developing ontolo-

gies. Using Ontology 101, an ontology development

methodology, the paper then describes SF-ONT - a

formal ontology that maps the knowledge domain of

solar irradiance forecasting. Finally, the paper des-

KEOD 2018 - 10th International Conference on Knowledge Engineering and Ontology Development

268

Figure 9: Class hierarchy relationship to identify class Mo-

delsForLSE.

Figure 10: The inferred class hierarchy of ModelsForLSE.

cribes the testing and validating of solar forecasting

ontology for accuracy and completeness using built-

in Prot ´eg ´e software reasoners.

SF-ONT is presented as a starting point and a ﬁrst

step for co-creating a shared ontological mapping of

the of the solar forecasting domain. Currently, SF-

ONT identiﬁes all the top level concepts in the dom-

ain of solar irradiance forecasting, and models their

relationships within the context of the current know-

ledge in the domain.

As research advances forecast models with bet-

ter spatial and temporal resolution, and as regula-

tory agencies continue to reﬁne metrics for compa-

ring forecast qualities, and as energy markets con-

tinue to evolve, SF-ONT will need to be updated

to reﬂect the most current state of domain know-

ledge. New instances will have to be populated

and existing relationships, properties and classes may

have to be (re)deﬁned. To facilitate reuse, dupli-

cation and updates, SF-ONT is made available at

http://pages.mtu.edu/ lebrown/research/sf-ont/.

As knowledge in the domain expands, this onto-

logy can be extended from the basic building box, mo-

diﬁed and maintained to suit the evolving needs of the

users. One priority area for SF-ONT extension is to

capture domain knowledge of PV systems planning

and design. For instance, concentrated solar plants

(CSP) and utility-scale solar farms (USF) are high

intensity generators directly connected to the trans-

mission infrastructure and dispatched through energy

markets, which calls for location speciﬁc forecasts at

a high temporal resolution. By contrast, rooftop solar

is connected to the distribution grid, often times ’be-

hind the meter’ with lower expectations for forecast

resolutions.

Another priority area for extension is PV system

security and cyber-security, with a particular empha-

sis on grid resiliency. The key challenge in this con-

text will be preserving interoperability between diffe-

rent domains of knowledge as their formal ontologies

are developed and mapped independently. In this con-

text, future updates to SF-ONT may need to actively

involve domain experts and end-users in the ontology

development process. Ontology engineering metho-

dologies like HCOME (Kotis and Vouros, 2006) em-

power domain experts and end users to ”manage on-

tologies shaping their common information space”,

which can lead to richer, more relevant, and regularly

updated ”living” ontologies.

REFERENCES

Barbieri, F., Rajakaruna, S., and Ghosh, A. (2017). Very

short-term photovoltaic power forecasting with cloud

modeling: A review. Renewable and Sustainable

Energy Reviews, 75:242–263.

Bonino, D., Corno, F., and De Russis, L. (2015). Power-

ont: An ontology-based approach for power consump-

tion estimation in smart homes. In Internet of Things.

User-Centric IoT, pages 3–8. Springer.

Cuenca, J., Larrinaga, F., and Curry, E. (2017). A uniﬁed

semantic ontology for energy management applicati-

ons. In WSP/WOMoCoE@ISWC.

Diagne, M., David, M., Lauret, P., Boland, J., and Schmutz,

N. (2013). Review of solar irradiance forecasting met-

hods and a proposition for small-scale insular grids.

Renewable and Sustainable Energy Reviews, 27:65–

76.

Esnaola-Gonzalez, I. and Bermudez, J. (2015). Energy

efﬁciency prediction semantic assistant onto-

logy. http://www.essepuntato.it/lode/owlapi/

https://w3id.org/eepsa.

FIESTA-IoT, E. H. (2017). Fiesta-ont. http://ontology.

ﬁesta-iot.eu/ontologyDocs/m3-lite.owl.

Gajderowicz, B. (2008). Weather-ont ontology. http://

www.scs.ryerson.ca/∼bgajdero/mscthesis/code/

ontologies/weather-ont-t2.owl.

Gruber, T. R. (1993). A translation approach to porta-

ble ontology speciﬁcations. Knowledge acquisition,

5(2):199–220.

Guarino, N., Oberle, D., and Staab, S. (2009). What is an

ontology? In Handbook on ontologies, pages 1–17.

Springer.

An Ontology for Solar Irradiation Forecast Models

269

Kotis, K. and Vouros, G. A. (2006). Human-centered on-

tology engineering: The hcome methodology. Know-

ledge and Information Systems, 10(1):109–131.

Maffei, A., Srinivasan, S., Castillejo, P., Mart

ınez, J. F.,

Iannelli, L., Bjerkan, E., and Glielmo, L. (2018). A

semantic-middleware-supported receding horizon op-

timal power ﬂow in energy grids. IEEE Transactions

on Industrial Informatics, 14(1):35–46.

McEvoy, A., Markvart, T., and Castaner, L. (2011). Practi-

cal Handbook of Photovoltaics, Second Edition: Fun-

damentals and Applications. Academic Press.

McGuinness, D. L., Van Harmelen, F., et al. (2004). Owl

web ontology language overview. W3C recommenda-

tion, 10(10):2004.

Musen, M. (2015). The prot

e project: A look back and a

look forward. AI Matters, 1(4):4–12.

Noy, N. F., McGuinness, D. L., et al. (2001). Ontology

development 101: A guide to creating your ﬁrst onto-

logy. Technical Report. Stanford University.

Paulescu, M. (2008). Solar irradiation via air temperature

data. In Modeling Solar Radiation at the Earths Sur-

face, pages 175–192. Springer.

Pelland, S., Remund, J., Kleissl, J., Oozeki, T., and De Bra-

bandere, K. (2013). Photovoltaic and solar forecas-

ting: state of the art. IEA PVPS, Task, 14.

Piazza, A. and Faso, G. (2014). Concentrated solar power:

Ontologies for solar radiation modeling and forecas-

ting. In Advances onto the Internet of Things, pages

325–337. Springer.

Prema, V. and Rao, K. U. (2015). Development of statistical

time series models for solar power prediction. Rene-

wable Energy, 83:100–109.

Reikard, G., Haupt, S. E., and Jensen, T. (2017). Fore-

casting ground-level irradiance over short horizons:

Time series, meteorological, and time-varying para-

meter models. Renewable Energy, 112:474–485.

Rijgersberg, H., van Assem, M., and Top, J. (2013). Onto-

logy of units of measure and related concepts. Seman-

tic Web, 4(1):3–13.

Ruiz-Arias, J. A. and Gueymard, C. A. (2018). Worldwide

inter-comparison of clear-sky solar radiation models:

Consensus-based review of direct and global irradi-

ance components simulated at the earth surface. Solar

Energy.

anchez-Cervantes, J. L., Radzimski, M., Rodriguez-

Enriquez, C. A., Alor-Hern

andez, G., Rodr

ıguez-

Mazahua, L., S

anchez-Ram

ırez, C., and Rodr

ıguez-

Gonz

alez, A. (2016). Sreqp: A solar radiation ex-

traction and query platform for the production and

consumption of linked data from weather stations sen-

sors. Journal of Sensors, 2016.

Staab, S. and Studer, R. (2013). Handbook on ontologies.

Springer Science & Business Media.

Staroch, P. (2013). A weather ontology for predictive cont-

rol in smart homes. MS Thesis.

Swartout, B., Patil, R., Knight, K., and Russ, T. (1996).

Toward distributed use of large-scale ontologies. In

Proc. of the Tenth Workshop on Knowledge Acquisi-

tion for Knowledge-Based Systems, pages 138–148.

Uschold, M. and Gruninger, M. (1996). Ontologies: Princi-

ples, methods and applications. The knowledge engi-

neering review, 11(02):93–136.

Uschold, M., King, M., Moralee, S., and Zorgios, Y. (1998).

The enterprise ontology. The knowledge engineering

review, 13(1):31–89.

Voyant, C., Notton, G., Kalogirou, S., Nivet, M.-L., Paoli,

C., Motte, F., and Fouilloy, A. (2017). Machine le-

arning methods for solar radiation forecasting: A re-

view. Renewable Energy, 105:569–582.

W3 (2003). Owl-geo. https://www.w3.org/2003/01/geo/.

W3 (2005). Weather-station ontology. https://www.w3.org/

2005/Incubator/ssn/ssnx/weather-station/station.

W3 (2017). Owl-time. https://www.w3.org/TR/owl-time/.

Widiss, R. and Porter, K. (2014). A review of variable gene-

ration forecasting in the west. NREL Technical Report,

303:275–3000.

Zieher, M., Lange, M., and Focken, M. (2015). Variable

renewable energy forecasting - integration into elec-

tricity grids and markets, a best practice guide. Ger-

man Federal Ministry for Economic Cooperation and

Development.

KEOD 2018 - 10th International Conference on Knowledge Engineering and Ontology Development

270