Multicriterion Topology Optimization Method and Automated Structural Optimization System

• Main Authors: Yu-hsin Chou, 鄒宇新
• Other Authors: none
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/41572153625656559594

博士 === 國立臺灣科技大學 === 機械工程系 === 97 === With the traditional topology optimization method, the large number of design variables often results in complicated design domains. The constraint functions often complicate the searching domain, making it difficult to find the global optimum and to increase the efficiency. Therefore, this dissertation proposes the multicriterion topology optimization method to tackle the above-mentioned disadvantages. Instead of using compliance as the objective function, the objective function of this new method includes both compliance and volume as two multicriterion functions to simplify the multimodality of the model and ease of attaining a consistent good result. Meanwhile, this new method also cancels all constraint functions, expanding the searching domain to increase the probability of getting global optimum and reducing the computation time to increase the efficiency. The resulting topological image obtained by the topology optimization method usually has indented interior holes and exterior boundary. Moreover, allowable stress or displacement is not under consideration, so the final configuration is not practical for direct application as a structural design. This research develops the image preprocessing technique, the image interpreting technique and the shape optimization variable defining technique to successfully integrate the topology optimization and the shape optimization into an autonomous procedure. Two automated structural optimization systems which are based on the characteristic value based image interpreting technique and the polygonal image interpreting technique are proposed. Two automated structural optimization systems are demonstrated and compared in several design examples.