An innovative deep neural network–based approach for internal cavity detection of timber columns using percussion sound

Abstract

Timber structures have been a dominant form of construction throughout most of history and continued to serve as a widely used staple of civil infrastructure in the modern era. As a natural material, wood is prone to termite damages, which often cause internal cavities for timber structures. Since internal cavities are invisible and greatly weaken structural load-bearing capacity, an effective method to timber internal cavity detection is of great importance to ensure structural safety. This article proposes an innovative deep neural network (DNN)–based approach for internal cavity detection of timber columns using percussion sound. The influence mechanism of percussion sound with the volume change of internal cavity was studied through theoretical and numerical analysis. A series of percussion tests on timber column specimens with different cavity volumes and environmental variations were conducted to validate the feasibility of the proposed DNN-based approach. Experimental results show high accuracy and generality for cavity severity identification regardless of percussion location, column section shape, and environmental effects, implying great potentials of the proposed approach as a fast tool for determining internal cavity of timber structures in field applications.