American Society of Civil Engineers


Building KBES for Diagnosing PC Pile With Inductive Learning


by Yi-Cherng Yeh, (Res. Asst., Dept. of Civ. Engrg., Nat. Cheng Kung Univ., Tainan, Taiwan, Republic of China), Yau-Hwaug Kuo, (Assoc. Prof., Inst. of Info. Engrg., Nat. Cheng Kung Univ., Tainan, Taiwan, Republic of China), and D. S. Hsu, (Prof., Dept. of Civ. Engrg., Nat. Cheng Kung Univ., Tainan, Taiwan, Republic of China)

Journal of Computing in Civil Engineering, Vol. 6, No. 2, April 1992, pp. 200-219, (doi:  http://dx.doi.org/10.1061/(ASCE)0887-3801(1992)6:2(200))

     Access full text
     Permissions for Reuse  

Document type: Journal Paper
Abstract: The damage of a prestressed concrete pile (PCP) during the driving process has resulted in injuries, time delay, and cost overruns. Diagnosing the damage is one of the most important problems in foundation engineering. A knowledge-based expert system (KBES) for diagnosing PCP is proposed in this paper. To overcome the glut of knowledge acquisition, the ID3 inductive learning algorithm is used to acquire knowledge rules. Five phases for building expert systems with inductive learning—identification, collection, implementation, refinement, and verification—are discussed. The knowledge base obtained from the inductive learning method is compared with that obtained from the conventional interview method in several aspects, including representation efficiency, reasoning efficiency, reasoning predictability, reasoning accuracy, and resources used. The results show that inductive learning is superior to the interview method in most aspects. The characteristics of civil engineering problems that make them good candidates for inductive learning are also discussed in this paper.


ASCE Subject Headings:
Algorithms
Assessment
Damage
Expert systems
Knowledge-based systems
Pile driving