Ž . Automation in Construction 8 1999 539–552 A neuro-fuzzy computational approach to constructability knowledge acquisition for construction technology evaluation Wen-der Yu a, ) , Miroslaw J. Skibniewski b a Department of Construction Engineering, Chung-Hua UniÕersity, Hsinchu 300, Taiwan b School of CiÕil Engineering, Purdue UniÕersity, West Lafayette, IN 47907-1294, USA Accepted 7 September 1998 Abstract This paper describes a methodology for constructability knowledge acquisition of construction technologies. The methodology combines a neuro-fuzzy network-based approach with genetic algorithms. The combination of fuzzy logic with learning abilities of neural networks and genetic algorithms may allow for automatic acquisition of constructability knowledge from training examples and for providing understandable explanations for the reasoning process. The proposed methodology can provide a mechanism to trace back factors causing unsatisfactory construction performance and the necessary feedback to construction engineers for technology innovation. An application example is provided to demonstrate the capabilities of the proposed methodology. q 1999 Published by Elsevier Science B.V. All rights reserved. Keywords: Construction technology evaluation; Constructability analysis; Neuro-fuzzy learning systems 1. Introduction In order to perform effective evaluation of a construction technology, the definition of technology performance should first of all be determined. The most important performance factors for a construc- tion technology include construction time, project cost, resource requirements, and product quality. Different construction project characteristics affect the performance of a specific technology. Thus, the performance of a specific construction technology is defined as the performance factor levels of the con- sidered technology working in specific construction project characteristics. In other words, the better the ) Corresponding author performance factor levels, the better the match be- tween the technology and the given project charac- teristics. The challenge therefore is how to optimize all performance factors. This issue can be regarded as a multi-objective optimization problem. More- over, each of the performance factors is influenced by many attributes. For example, the construction time is influenced by task volume, availability of labor and equipment, weather conditions, etc. Opti- mizing one performance factor usually conflicts with similar attempts regarding the others. For example, achieving shorter construction time requires more Ž . resources ‘Crash time’ mode , increasing the project cost. Such relationships among performance factors and their influencing factors are difficult, if not impossible, to express for experienced engineers. 0926-5805r99r$ - see front matter q 1999 Published by Elsevier Science B.V. All rights reserved. Ž . PII: S0926-5805 98 00104-6