Search

Planning of construction technical specifications (CTS) for deep foundations is critical for ensuring works performed safely. k-nearest neighbors (kNN) is regarded as a practical algorithm for case retrieval in a case-based reasoning (CBR) cycle to search for past similar plans for new plan making. The parameter k and neighbors' weights affect the performance of the CBR cycle deeply but kNN neglects the weights' effect on case retrieval. The massive and multisource data of CTS of deep foundations presents a challenge for retaining case data in a database and for decision making due to an inefficient data process of the traditional tool. This paper presents a new framework to integrate weighted k-nearest neighbors (kkNN) to improve the performance of a CBR system for technical planning of deep foundations. It contains two parts: (1) a process to deal with a large amount of data derived from CTS; and (2) kkNN to obtain similar cases considering k and the weights of neighbors'. The feasibility of the proposed approach is validated through a case study and the evaluation result shows that the approach enhances the performance of the CBR cycle in creating construction technical specifications in deep foundation projects. [ABSTRACT FROM AUTHOR]