Pregled bibliografske jedinice broj: 1217494
Evolutionary Numerical Workflow for Large-Scale Feature-based Remodeling and Shape Optimization
Evolutionary Numerical Workflow for Large-Scale Feature-based Remodeling and Shape Optimization // Machine Learning and Data Mining in Pattern Recognition / Perner, Petra (ur.).
Leipzig: ibai-publishing, 2022. str. 167-178 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 1217494 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Evolutionary Numerical Workflow for Large-Scale
Feature-based Remodeling and Shape Optimization
Autori
Vučina, Damir ; Ćurković, Milan ; Marinić-Kragić, Ivo
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Izvornik
Machine Learning and Data Mining in Pattern Recognition
/ Perner, Petra - Leipzig : Ibai-publishing, 2022, 167-178
ISBN
978-3-942952-93-4
Skup
18th International Conference on Machine Learning and Data Mining (MLDM 2022)
Mjesto i datum
New York City (NY), Sjedinjene Američke Države, 16.07.2022. - 21.07.2022
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
Evolutionary Shape Optimization ; Efficient 3D Parameterization ; Intelligent Numerical Workflow ; Adaptive Geometric Modeling ; 3D feature recognition ; Partitions
Sažetak
Evolutionary parametric shape optimization relies on generic shape models which provide sufficient 3D geometric modeling freedom while being modest in terms of the number of shape variables and hence dimensionality of search space. The number of shape parameters needs to be reduced towards computational efficiency in shape optimization. However, this may imply loss of generality or local modeling capacity, potentially also bias towards unintentionally predefined 3D shape templates, all of which may result in sub-optimal shapes. This paper develops a novel approach in the form of a conceptual intelligent system based on integral parametric surfaces representing objects. The computational efficiency of the models is established by engaging a sparing multitude of shape parameters in evolutionary optimization. Sufficiently generic and unbiased modeling capacity is obtained by the adaptability of the approach, as it can adapt to the features of the current state in optimization iterations. Decomposition of the integral surfaces into shape partitions is implemented based on 3D pattern recognition (edges, peaks, etc).
Izvorni jezik
Engleski
Znanstvena područja
Strojarstvo, Temeljne tehničke znanosti, Interdisciplinarne tehničke znanosti
POVEZANOST RADA
Projekti:
IP-2018-01-6774 - EVOLUCIJSKA SINTEZA OBLIKA S INTEGRALNIM I PARTICIONIRANIM 3D FENOTIPOVIMA, DINAMIČKOM PARAMETRIZACIJOM I BEZMREŽNIM MODELIRANJEM (Synth3D) (Vučina, Damir, HRZZ - 2018-01) ( CroRIS)
Ustanove:
Fakultet elektrotehnike, strojarstva i brodogradnje, Split
