The development of distributed particle swarm optimization and its application on system identification

• Main Authors: King, Uei-Rung, 金瑋榮
• Other Authors: Hung, Shih-Lin
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/6b8pbe

碩士 === 國立交通大學 === 土木工程系所 === 107 === Structures that have severe vibrations effects may suffer structural damage. The damage will deteriorate the structure if it is not detected and repaired in time. Therefore, system identification is required to detect the health of the structure. This research applied MPI technology and cluster system to develop distributed particle swarm optimization which is written by Python. It might solve the problem of long runtime on the high dimensional problem by traditional particle swarm optimization. Although the least squares method is the most widely-used approach in solving the parameters in ARX model, this study intends to use particle swarm optimization to solve the parameters in ARX model. The feasibility of this method is validated by three-story steel structure, five-story steel structure and malfunctional five-story steel structure.