基于GP的创新设计技术研究及应用

【作者】 江海燕；

【导师】 刘弘；

【作者基本信息】 山东师范大学 ， 计算机软件与理论， 2006， 硕士

【摘要】 J.R.Koza教授在1989年提出了GP(Genetic Programming)算法。GP最重要的特点之一是组成种群的个体有一种动态的树状结构。GP的基本遗传算子包括选择、交叉和变异。作为进化计算的较新分支之一,GP受到了各领域专家的重视。自1994年以来,IEEE进化计算国际会议就设置了“GP”的专题,对GP的算法、理论、数据结构等方面进行研究。GP试图解决计算科学的一个中心问题:计算机在没有明显编程的情况下怎样解决问题。直到现在,GP在设计、优化控制、系统识别、函数学习、计算分子生物学等诸多问题领域中得到了广泛和深入的应用,体现了GP的发展优势及突出特点。设计在经历了萌芽阶段,手工艺阶段,到达了现代工业设计阶段。现代设计是一种创造性活动,它要求考虑计算机技术的应用,综合功能、技术、材料、经济、安全、审美等因素进行规划与设计,以使设计的产品与产品的使用者之间取得最佳的匹配。创新设计作为适应现今社会发展需求而产生的一种现代设计方法,已经得到诸多设计人员的重视,其最本质的特点是创新。创新意味着需要更多的跨学科的知识结构、更复杂的技术支撑和更完善的创新理论。在后工业化时代,人们把产品的设计创新称为继科技创新之后的“第二次竞争”。将进化计算应用于创新设计,在国内外都有一定的研究与实现。进化计算应用于建筑设计、工业设计、艺术设计、服装设计、图像修补和音乐创作等领域已经取得了一定的成绩。将GP的相关理论和技术应用于创新设计领域,将进化算法与设计师以往的设计经验相结合,综合利用计算机技术、人工智能技术得到新颖而能满足设计要求的创新结果,是可行且又有深远意义的。本文所做主要工作如下:1、对GP理论进行研究。将GP与其它相关理论进行比较,得到将GP应用于创新设计的优势所在。2、针对传统设计方法存在的问题,以及现代设计的要求,指明创新设计的重要性。文章对创新设计的概念,创新设计的方法手段及创新设计的模型进行了探讨和研究。3、根据将进化计算应用于创新设计的研究背景和现状,文章将GP应用于创新设计,并体现在建筑设计的应用上。在建筑设计中建筑构件的生成主要是采用GP算法,实现由计算机自动产生具有创新外观造型的建筑构件。并且,为了激发设计师的创作灵感,利用群体智能之PSO算法,将得到的创新的优良的建筑构件进行智能组装,从而得到较优秀的建筑体。本文研究受国家自然科学基金《支持创新概念设计的协同进化计算技术研究》(项目号:60374054)、山东省自然科学基金项目《支持创新设计的进化计算技术研究》(项目号:Y2003G01)、山东省中青年科学家奖励基金项目《创新设计软构件及集成系统》(项目号:03BS003)等项目资金的资助。更多还原

【Abstract】 Professor J.R.Koza proposed the algorithm of GP (Genetic Programming ) in 1989. Themost important character of GP is that the individual in the population has the data structure ofTree. The basic genetic operators of GP are mutation, crossover, and selection. As one of the newbranches of Evolutionary Computation, GP is valued by experts in all kinds of fields. Since 1994,the international evolutionary computation conference of IEEE has established the "GP" topic,which is to research GP algorithm, theory, and data structure and so on. GP attempts to solve acrucial question in computation science: How does the computer solve a problem withoutobvious programming situation. Till now, GP has been applied extensively and deeply in thefields of design, optimum control, system recognition, function learning, and calculationmolecular biology and so on. All these have reflected the developing advantages and outstandingcharacteristics of GP.And after the courses of burgeoning stage and handicraft stage in design, it has reached theperiod of modern industrial design. Modern design requests the synthetic applications ofcomputer technology. It synthesizes the factors of function, technology, material, economy, safetyand beauty and so on. It is a creative activity, which is to try to get the best matches between theproducts of design and the consumers of the products. The most essential characteristic ofCreative Design, which is a modern design method, accepted by more and more designers andadapts to the demands of the development of the society, is innovation. The innovation ofproducts means that more interdisciplinary knowledge structures, more complicated technicalsupports and more complete creative theory are needed. In the period of then industry era, thecreative design of products is called “the second competition” after scientific and technicalinnovation.EC has been applied to Creative Design in and abroad. Using EC in the fields of architecturaldesign, industrial design, art design, costume design, image repairment and music has come true,and some systems have been realized.It has far-reach meaning to apply GP to the field of Creative Design, to combine thealgorithms of Evolutionary Computation and the past experience of the designers’, and to utilizethe computer technology, artificial intelligence, and so on synthetically to get the creative results,which meet the requests of the designers. The main tasks that this paper has done are as follows:1.To research the theory of GP(Genetic Programming). Comparing the theory of GP withother relevant theories to get the advantages of applying GP in the Creative Design.2.To point out the importance of Creative Design, according to the shortcomings that thetraditional design method exists, and the demands of modern design. It has studied the concept ofCreative Design, the methods and means for Creative Design and also the models of CreativeDesign in this paper.3.According to the background and current situation of applying the algorithms ofEvolutionary Computation to Creative Design, this paper tries to apply the algorithm of GP toCreative Design and tries to realize the application in architectural design. to generate creativearchitectural components automatically in computer with the usage of GP algorithm. And, inorder to stimulate the designers’ inspirations, the algorithm of PSO, as one of the SwarmIntelligence Algorithms, can be used, which can get the more outstanding building bodies byintelligent assembling.更多还原