Member separation and deformation recognition of spatial grid structures in-service

Abstract

Bending deformation of members in spatial grid structures is a typical damage threatening structural safety. Conventional methods of deformation measurement have difficulty in detection member selection, high-altitude operation, or systematic error elimination. And reverse modeling technology for deformation detection is mostly utilized in entire structures or laboratory scenes. In this study, two kinds of member separation algorithms are developed based on point cloud registration and point cloud features respectively, both of which are suitable for engineering situations of insufficient point cloud density and multiple noise sources. The deformation recognition algorithm for members is improved accordingly with double judgment conditions. A field test has been carried out in a Sports Center Stadium in Guangzhou province. 7045 members are identified, and 29 crooked members are eventually recognized. The results demonstrate that the proposed algorithms can improve the efficiency of member separation and deformation recognition, and assist the monitoring and assessment of spatial grid structures in-service.