Nature-Inspired Metaheuristic Support Vector Regression System for Civil Engineering Managers

• Main Author: Dac-Khuong Bui
• Other Authors: Jui-Sheng Chou
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/09575489184597797053

碩士 === 國立臺灣科技大學 === 營建工程系 === 103 === Developing an expert system has been considered as complex and knowledge driven process. This study proposes a nature-inspired metaheuristic regression system that can find appropriate solutions. The system uses a graphical user interface but does not require a mathematical program installation. The user-friendly interface was designed in Matlab GUIDE and was implemented by Matlab compiler. The standalone system is easy to use and has many functions, including evaluation: use opened data file, select test set, hold-out, cross validation and prediction to solve many civil engineering problems with simple manipulations on the interface of system. Five benchmark functions were used to evaluate the effectiveness of the optimization approach. The performance of proposed system was then validated by comparing its solutions obtained for civil engineering problems with those obtained by empirical methods reported previously. Five actual data sets including energy-efficient buildings, construction material strength, concrete structure shear strength, bridge scour depth, and sub base soil modulus were used as case studies. The prediction accuracy were 8.24% – 91.76% better than those of previously reported models. The analytical results support the feasibility of using the proposed system to solve civil engineering problems. The system was also much faster at identifying the optimum parameters and solving problems. The experiments confirmed that the novel nature-inspired metaheuristic regression system proposed in this study has superior efficiency, effectiveness, and accuracy.