Priložnosti generativnega pristopa k načrtovanju stavb

magistrsko delo

Luka Gradišar (Author), Žiga Turk (Mentor), Robert Klinc (Co-mentor)

Abstract

Gradbeništvo uporablja številne računalniške tehnologije. Pretežno gre za tehnologije, ki analizirajo že zamišljene rešitve in take, ki pomagajo pri predstavitvi in dokumentiranju zamisli. Nedavni razvoj umetne inteligence in sorodnih področij, pa odpira možnosti, da %računalnik% rešitve tudi bolj ali manj samostojno kreira. Eden od takih pristopov je metoda generativnega načrtovanja. V nalogi bodo predstavljene teoretične osnove te metode, potem pa bo aplicirana na primeru projektiranja elementov za senčenje stavbe v okolju programa Revit. Realen inženirski problem energetsko učinkovitega senčenja je bil poenostavljen in prilagojen v obliko zapisa za programsko okolje Dynamo. Predstavljen bo način vizualnega programiranja v tem okolju. Optimalna rešitev se kreira tako, da izberemo in definiramo vhodne spremenljivke ter kriterije, ki podajo numerično oceno alternativnim rešitvam. Po principu generativnega pristopa genetski algoritmi generirajo rešitve, ki jih testirajo in sortirajo ter se z vsako novo iteracijo bližajo optimalnim vrednostim. Generirajo in analizirajo prostor rešitev, da bi lahko potem avtonomno ali s posredovanjem inženirja izbrali končno rešitev. V nalogi postopek primerjamo s tradicionalnim načrtovalskim pristopom ter ocenimo prednosti in slabosti generativnega pristopa. Glavna ideja pristopa je sodelovanje projektanta z inteligentnim orodjem, ki ima komplementarne sposobnosti: pregled velikega števila podatkov, generiranje velikega števila alternativ, analiza rešitev, inkrementalno izboljšanje rešitev. Pristop je splošen in zato uporaben tudi kot pomoč pri načrtovanju na drugih področjih gradbeništva.

Keywords

gradbeništvo;magistrska dela;II. GR;Generativni pristop;računalniško podprto modeliranje;optimizacija;genetski algoritmi;BIM;parametrično modeliranje;Dynamo;Project Refinery;vizualno programiranje;

Data

Language:	Slovenian
Year of publishing:	2019
Typology:	2.09 - Master's Thesis
Organization:	UL FGG - Faculty of Civil and Geodetic Engineering
Publisher:	[L. Gradišar]
UDC:	004.42:69(497.4)(043.3)
COBISS:	8886369
Views:	1271
Downloads:	320
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Generative design opportunities to the building design
Secondary abstract:	Civil engineering field uses many different computer technologies. Mainly these are technologies for analysis of the already defined ideas and those which help with the presentation and documentation of said ideas. However, recent development in the artificial intelligence and related fields opens the possibility for the computer to independently generate solutions. One such approach is the generative design. The thesis outlines the theoretical basis of this method, which is then applied to the practical example of the shading elements for the building in Revit. Real engineering example of the energy efficient shading has been simplified and conformed to the Dynamo programming environment in which the visual programming will be presented. In order to find the optimal solution, it is necessary to select and define the input variables and objectives with fitness functions. In the generative design process, the solutions are generated, tested and sorted by genetic algorithms, and with each new iteration we get closer to the optimal solution. They generate and analyze the solution space, to find the final solution autonomously or with the help of an engineer. We compare this process with the design approach and outline the pros and cons of the generative approach. The main idea of the generative design is the collaboration between the designer and the machine, which has complementary capabilities, such as: sorting large amount of data, generating large number of alternatives, analysis of the solutions and iterative improvement of the solution. The approach is general and can be used to help with the designs in other areas of civil engineering.
Secondary keywords:	civil engineering;master thesis;Generative design;computational design;optimization;BIM;parametric modeling;Dynamo;Project Refinery;visual programing;genetic algorithms;
Type (COBISS):	Master's thesis/paper
Study programme:	0
Embargo end date (OpenAIRE):	1970-01-01
Thesis comment:	Univ. v Ljubljani, Fak. za gradbeništvo in geodezijo
Pages:	XVIII, 51 str.
ID:	11222821