Three level modelling of uncertainties in the condition assessment of buildings

Ádám Bukovics, István Harmati, L. Kóczy

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Condition survey evaluation analyses (enginering-static expert reports) are often prepared about building structures (often focusing on load-bearing structures) in order to support appropriate maintenance and repair of the existing buildings. Based on these analyses modelling of the condition of buildings and building structures can be a big help to adopt decisions on intervention. In the course of our former research condition evaluation, decision support and ranking method was worked out which, based on a unified system of viewpoints, considering various priorities, is able to determine the condition of residential buildings. We have used tree-structure, fuzzy singleton signature based model, in which the expert specifies one discrete value for the condition of every examined building structure. In the course of our research it was experienced that the elaborated method is too subjective and uncertain, therefore the method is being further developed so that modelling of objective and subjective uncertainties become possible in the course of preparing expert opinions. Three levels were created to model uncertainties. Uncertainty on level 1 is how to transform verbal evaluations into fuzzy membership functions (verbal values cannot be unambiguously transformed into numerical values). In this model instead of membership values, which were used in the former model, we assign linguistic label modelling membership functions to the leaves of the structure. Level 2 we are modelling an uncertainty where even the expert is unable to precisely determine the condition of the examined structure. Often an interval is specified for the condition of the structure instead of a specific status value. It means that the condition of structure can have any value between two specified values with the same probability. To model it the triangular-shaped membership function is transformed into a trapezoidal-shaped membership function with the help of linguistic hedges. On level 3 it is modelled that the expert evaluation itself is not considered totally reliable. Subjectivity, professional preparedness of the expert, conducting the examination, or the quality of available circumstances and data may significantly influence the reliability of the final result. These uncertainties can be modelled by modifying the shapes of the membership function. The fuzzy set signature based model created by modelling the uncertainties at three levels is able to model with proper accuracy (more sophistically as compared to the former model) the condition of buildings, which were specified by verbal evaluation in expert opinion.

Original languageEnglish
Title of host publicationISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications
PublisherFuji Technology Press
ISBN (Electronic)9784990534349
Publication statusPublished - 2016
Event7th International Symposium on Computational Intelligence and Industrial Applications, ISCIIA 2016 - Beijing, China
Duration: Nov 3 2016Nov 6 2016

Other

Other7th International Symposium on Computational Intelligence and Industrial Applications, ISCIIA 2016
CountryChina
CityBeijing
Period11/3/1611/6/16

Fingerprint

Membership functions
Linguistics
Bearings (structural)
Uncertainty
Fuzzy sets
Labels
Repair

Keywords

  • Fuzzy Logic
  • Linguistic Hedges
  • Membership Functions
  • Residential Building

ASJC Scopus subject areas

  • Artificial Intelligence
  • Industrial and Manufacturing Engineering

Cite this

Bukovics, Á., Harmati, I., & Kóczy, L. (2016). Three level modelling of uncertainties in the condition assessment of buildings. In ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications Fuji Technology Press.

Three level modelling of uncertainties in the condition assessment of buildings. / Bukovics, Ádám; Harmati, István; Kóczy, L.

ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications. Fuji Technology Press, 2016.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Bukovics, Á, Harmati, I & Kóczy, L 2016, Three level modelling of uncertainties in the condition assessment of buildings. in ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications. Fuji Technology Press, 7th International Symposium on Computational Intelligence and Industrial Applications, ISCIIA 2016, Beijing, China, 11/3/16.
Bukovics Á, Harmati I, Kóczy L. Three level modelling of uncertainties in the condition assessment of buildings. In ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications. Fuji Technology Press. 2016
Bukovics, Ádám ; Harmati, István ; Kóczy, L. / Three level modelling of uncertainties in the condition assessment of buildings. ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications. Fuji Technology Press, 2016.
@inproceedings{1f5035899c8646f8a6c945af6c573877,
title = "Three level modelling of uncertainties in the condition assessment of buildings",
abstract = "Condition survey evaluation analyses (enginering-static expert reports) are often prepared about building structures (often focusing on load-bearing structures) in order to support appropriate maintenance and repair of the existing buildings. Based on these analyses modelling of the condition of buildings and building structures can be a big help to adopt decisions on intervention. In the course of our former research condition evaluation, decision support and ranking method was worked out which, based on a unified system of viewpoints, considering various priorities, is able to determine the condition of residential buildings. We have used tree-structure, fuzzy singleton signature based model, in which the expert specifies one discrete value for the condition of every examined building structure. In the course of our research it was experienced that the elaborated method is too subjective and uncertain, therefore the method is being further developed so that modelling of objective and subjective uncertainties become possible in the course of preparing expert opinions. Three levels were created to model uncertainties. Uncertainty on level 1 is how to transform verbal evaluations into fuzzy membership functions (verbal values cannot be unambiguously transformed into numerical values). In this model instead of membership values, which were used in the former model, we assign linguistic label modelling membership functions to the leaves of the structure. Level 2 we are modelling an uncertainty where even the expert is unable to precisely determine the condition of the examined structure. Often an interval is specified for the condition of the structure instead of a specific status value. It means that the condition of structure can have any value between two specified values with the same probability. To model it the triangular-shaped membership function is transformed into a trapezoidal-shaped membership function with the help of linguistic hedges. On level 3 it is modelled that the expert evaluation itself is not considered totally reliable. Subjectivity, professional preparedness of the expert, conducting the examination, or the quality of available circumstances and data may significantly influence the reliability of the final result. These uncertainties can be modelled by modifying the shapes of the membership function. The fuzzy set signature based model created by modelling the uncertainties at three levels is able to model with proper accuracy (more sophistically as compared to the former model) the condition of buildings, which were specified by verbal evaluation in expert opinion.",
keywords = "Fuzzy Logic, Linguistic Hedges, Membership Functions, Residential Building",
author = "{\'A}d{\'a}m Bukovics and Istv{\'a}n Harmati and L. K{\'o}czy",
year = "2016",
language = "English",
booktitle = "ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications",
publisher = "Fuji Technology Press",
address = "Japan",

}

TY - GEN

T1 - Three level modelling of uncertainties in the condition assessment of buildings

AU - Bukovics, Ádám

AU - Harmati, István

AU - Kóczy, L.

PY - 2016

Y1 - 2016

N2 - Condition survey evaluation analyses (enginering-static expert reports) are often prepared about building structures (often focusing on load-bearing structures) in order to support appropriate maintenance and repair of the existing buildings. Based on these analyses modelling of the condition of buildings and building structures can be a big help to adopt decisions on intervention. In the course of our former research condition evaluation, decision support and ranking method was worked out which, based on a unified system of viewpoints, considering various priorities, is able to determine the condition of residential buildings. We have used tree-structure, fuzzy singleton signature based model, in which the expert specifies one discrete value for the condition of every examined building structure. In the course of our research it was experienced that the elaborated method is too subjective and uncertain, therefore the method is being further developed so that modelling of objective and subjective uncertainties become possible in the course of preparing expert opinions. Three levels were created to model uncertainties. Uncertainty on level 1 is how to transform verbal evaluations into fuzzy membership functions (verbal values cannot be unambiguously transformed into numerical values). In this model instead of membership values, which were used in the former model, we assign linguistic label modelling membership functions to the leaves of the structure. Level 2 we are modelling an uncertainty where even the expert is unable to precisely determine the condition of the examined structure. Often an interval is specified for the condition of the structure instead of a specific status value. It means that the condition of structure can have any value between two specified values with the same probability. To model it the triangular-shaped membership function is transformed into a trapezoidal-shaped membership function with the help of linguistic hedges. On level 3 it is modelled that the expert evaluation itself is not considered totally reliable. Subjectivity, professional preparedness of the expert, conducting the examination, or the quality of available circumstances and data may significantly influence the reliability of the final result. These uncertainties can be modelled by modifying the shapes of the membership function. The fuzzy set signature based model created by modelling the uncertainties at three levels is able to model with proper accuracy (more sophistically as compared to the former model) the condition of buildings, which were specified by verbal evaluation in expert opinion.

AB - Condition survey evaluation analyses (enginering-static expert reports) are often prepared about building structures (often focusing on load-bearing structures) in order to support appropriate maintenance and repair of the existing buildings. Based on these analyses modelling of the condition of buildings and building structures can be a big help to adopt decisions on intervention. In the course of our former research condition evaluation, decision support and ranking method was worked out which, based on a unified system of viewpoints, considering various priorities, is able to determine the condition of residential buildings. We have used tree-structure, fuzzy singleton signature based model, in which the expert specifies one discrete value for the condition of every examined building structure. In the course of our research it was experienced that the elaborated method is too subjective and uncertain, therefore the method is being further developed so that modelling of objective and subjective uncertainties become possible in the course of preparing expert opinions. Three levels were created to model uncertainties. Uncertainty on level 1 is how to transform verbal evaluations into fuzzy membership functions (verbal values cannot be unambiguously transformed into numerical values). In this model instead of membership values, which were used in the former model, we assign linguistic label modelling membership functions to the leaves of the structure. Level 2 we are modelling an uncertainty where even the expert is unable to precisely determine the condition of the examined structure. Often an interval is specified for the condition of the structure instead of a specific status value. It means that the condition of structure can have any value between two specified values with the same probability. To model it the triangular-shaped membership function is transformed into a trapezoidal-shaped membership function with the help of linguistic hedges. On level 3 it is modelled that the expert evaluation itself is not considered totally reliable. Subjectivity, professional preparedness of the expert, conducting the examination, or the quality of available circumstances and data may significantly influence the reliability of the final result. These uncertainties can be modelled by modifying the shapes of the membership function. The fuzzy set signature based model created by modelling the uncertainties at three levels is able to model with proper accuracy (more sophistically as compared to the former model) the condition of buildings, which were specified by verbal evaluation in expert opinion.

KW - Fuzzy Logic

KW - Linguistic Hedges

KW - Membership Functions

KW - Residential Building

UR - http://www.scopus.com/inward/record.url?scp=84997707874&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84997707874&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84997707874

BT - ISCIIA 2016 - 7th International Symposium on Computational Intelligence and Industrial Applications

PB - Fuji Technology Press

ER -