الگوریتم تکاملی ترکیبی برای میزان سازی یک مدل شبیه سازی از پارچه
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|9725||2012||8 صفحه PDF||سفارش دهید||5254 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Applied Soft Computing, Volume 12, Issue 1, January 2012, Pages 266–273
Textile simulation models are notorious for being difficult to tune. The physically based derivations of energy functions, as mostly used for mapping the characteristics of real-world textiles on to simulation models, are labour-intensive and not guarantee satisfactory results. The extremely complex behaviour of textiles requires additional adjustment over a wide-range of parameters in order to achieve realistic real-life behaviour of the model. Furthermore, such derivations might not even be possible when dealing with mass-spring particle system-based models. Since there is no explicit correlation between the physical characteristics of textiles and the stiffnesses of springs that control a model’s behaviour, this remains an unresolved issue. This paper proposes a hybrid evolutionary algorithm (EA), in order to solve this problem. The initial parameters of the model are written in individual’s genes, where the number of genes is predefined for different textile types in order to limit the search-space. By mimicking the evolution processes, the EA is used to search the stability domain of the model to find a set of parameters that persuasively imitate the behaviour of a given real-world textile (e.g. silk, cotton or wool). This evaluation is based on the drape measurement, a characteristic often used when evaluating fabrics within the textile industry. The proposed EA is multi-objective, as textile drape is analysed using different quantifications. Local search is used to heuristically improve convergence towards a solution, while the efficiency of the method is demonstrated in comparison to a simple EA. To the best of our knowledge, this problem is being solved using an EA for the first time.
Considerable research efforts have been dedicated to recreating the realistic appearances and behaviours of cloths for computer simulation. Early computer models of textiles (such as , ,  and ) are based on geometric constrains . They are suitable for generating the static shapes of textiles and can produce the realistic appearances of textiles’ folds and creases. However, classical geometric representations  cannot incorporate the physical properties of textiles. Therefore, the realistic behaviours of cloths cannot be obtained and dynamic physically based models have been needed. One of the earliest physically based models was proposed by Feynman . This model applies the theory of elastic shells as the energy functions employed to define the behaviours of the shells. Free form surface models of cloths are proposed by Trezopoulus et al. in ,  and . These models utilize a system of differential equations derived from the elastic theory, in order to represent the energy functions of surface deformations. Discretization is achieved either by using a finite difference method or a finite element method, while the solution to the system of equations is found by numerical integration. Further development of Trezopoulus’ models is reported by Carignan, who considered self-collision detection in cloth simulation . Continuous models have strong theoretical backgrounds, yet they produce unsatisfying simulation results. The main reason is in the relatively coarse structures of the textiles  and . Various discrete physically based models have been proposed because of this. The approach based on a particle-system, as presented in , uses the Kawabata evaluation system (KES)  in order to obtain the data needed for the derivations of energy functions. Another particle system, often used in contemporary models, is the mass-spring particle system , ,  and . Each particle has a mass and is connected to neighbouring particles by stretching, bending, and shearing springs . These models are computationally efficient, simple to implement, and achieve good results in real-time applications ,  and . Different methods exist in regard to how formal particle systems  are described. One of the most efficient is the so-called Verlet integration . Our application uses its optimization as described by Jakobsen . In this way, the real-time behaviour of the system with self-collision detection has been obtained  on an ordinary PC.
نتیجه گیری انگلیسی
Mapping the physical characteristics of textiles to the parameters needed in textile simulation models, is an extremely complex task for the end-users of textile modelling tools. An EA has been proposed for mapping the characteristics of those textiles obtained from a drapemeter, into a mass-spring particle simulation system. The initial SEA was not a viable solution due to its slow convergence rate. Hence, our primary goal was to improve algorithm’s convergence. This paper proposed an efficient hybrid EA for tuning a textile simulation model. The effectiveness of the hybrid EA is derived from the use of a local search operator, where searching is heuristically directed towards the local optimum within the neighbourhood of the individual. However, local search is a computationally expensive operation and is inefficient when performed on each and every individual. Rather than this, only the best individual in the population should be used for this purpose. Additional improvements have been made by limiting the search-space with the heuristic elimination of springs (and related genes) from the model, and by decreasing the mutation step over generations. In this way, convergence towards a solution is highly increased in comparison with the SEA, as confirmed by results. Using the hybrid EA, it is possible to achieve realistic real-life behaviour for textile simulation models, which could be used for cloth behaviour studies in interactive animations and CAD systems. These models can be observed at http://gemma.uni-mb.si/textile/index.html.