Analysis of collaborative CAD modelling activities using machine learning methods

master's thesis

• Title (croatian): Analiza aktivnosti kolaborativnog CAD modeliranja upotrebom metoda strojnog učenja
• Author: Robert Celjak
• Mentor: Stanko Škec (mentor)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2023-05-11, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Mechanical Engineering

Abstract

Computer-Aided Design (CAD) models are one of the primary artefacts in the design process. They allow designers to develop and collaborate on design within the design workflow. The advent of cloud computing has enhanced collaboration in design by enabling simultaneous real-time work on shared CAD models. To determine the best approaches toward design team composition and users’ modelling practises, it is crucial to identify user archetypes and designers’ sequential patterns across different aspects of the CAD modelling process. This thesis presents the results of the implementation of clustering and pattern recognition machine learning methods on a dataset obtained from a project-based design course. The data was collected in a non-intrusive manner, pre-processed, and organised within an adopted and modified CAD action classification. The data analysis was performed on a dataset which included 14 three-member design teams, which performed 547 357 CAD actions within the design project. The results show that higher-performing teams worked individually while lower-performing worked individually and as a team equally. Lower-performing teams had role overlap among designers, while higher-performing teams had clearly defined roles for part, assembly, and versatile designers. Two team compositions were observed based on the types of CAD actions performed. The first had two or all three members performing similar amounts of creating, editing, and organising classes of CAD actions. The second had a dominant, versatile user leading the team in actions, related to said classes, performed. The pattern recognition algorithm identified common sequential patterns in the modelling processes of users. These results can lead to the identification of user archetypes which can facilitate forming design teams of complementary members. These results provide valuable insights for engineers and educators seeking to understand the sequential patterns of the modelling process and their potential impact on CAD model quality.

Abstract (croatian)

CAD (eng. Computer-Aided Design) modeli su jedan od glavnih artefakata u procesu konstruiranja. Inženjerima omogućuju da razvijaju dijelove proizvoda te surađuju tijekom procesa razvoja. Pojava računalstva u oblaku (eng. Cloud computing) je unaprijedila suradnju omogućujući istovremeni rad na dijeljenim CAD modelima. S ciljem određivanja optimalnog sastava razvojnih timova te pristupa CAD modeliranju, ključno je identificirati korisničke arhetipove i sekvencijalne obrasce u različitim aspektima procesa CAD modeliranja. Ovaj rad predstavlja rezultate implementacije metoda strojnog učenja – klasteriranja te prepoznavanja obrazaca na skupu podataka prikupljenom tijekom izvođenja CAD projektnog kolegija. Podaci su prikupljeni na neinvazivan način, pripremljeni su za analizu podataka te organizirani unutar klasifikacije CAD aktivnosti. Analiza podataka provedena je na setu podataka koji je uključivao 14 tročlanih razvojnih timova koji su izveli 547 357 CAD akcija tijekom razvojnog projekta. Rezultati pokazuju da su bolji timovi radili individualno, dok su lošiji timovi podjednako radili individualno i u timu. Sastav lošijih timova karakteriziran je preklapanjem timskih uloga, dok su bolji timovi imali jasno definirane uloge članova tima (part, assembly i versatile designer). Uočene su dvije vrste sastava timova u kontekstu korištenih vrsta CAD aktivnosti. Prva vrsta uključivala je dva ili sva tri člana koji su podjednako koristili klase generiranja, modificiranja i organiziranja CAD aktivnosti. Druga vrsta je imala dominantnog, raznovrsnog člana koji je predvodio tim u korištenju CAD aktivnosti navedenih klasa. Algoritam prepoznavanja obrazaca identificirao je frekventne sekvencijalne obrasce u procesu CAD modeliranja. Ovi rezultati vode do identifikacije korisničkih arhetipova koji mogu potpomognuti proces formiranja razvojnih timova s komplementarnim članovima. Navedeni rezultati pružaju vrijedne uvide inženjerima i istraživačima koji žele razumjeti sekvencijalne obrasce procesa CAD modeliranja i njihov potencijalni utjecaj na konačnu kvalitetu CAD modela.

• Keywords: CAD modelling; machine learning; user archetypes; clustering; pattern recognition; collaboration
• Keywords (croatian): CAD modeliranje; strojno učenje; korisnički arhetipovi; klasteriranje; identifikacija obrazaca; kolaboracija
• Language: english
• URN:NBN: urn:nbn:hr:235:141686
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka strojarstva (magistar/magistra inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Embargoed access; Embargo expiration date: 2025-05-08
• Created on: 2023-05-08 18:31:32