Métodos de agrupamento longitudinal: uma aplicação em dados de ondas ultrassônicas

Abstract

The preservation of buildings faces several challenges, such as the wear caused by weather, atmospheric conditions, soil settlement or even lack of proper maintenance. In some cases, such as historical heritage, it is not possible to carry out traditional destructive tests to verify the physical and mechanical quality of materials and structure. Then, in order to preserve the structure of buildings, non-destructive tests are a great advance in civil and mechanical engineering and these tests using ultrasonic waves, in particular, have been shown to be efficient in providing necessary information to detect damage in the interior of masonry and, so, predict situations. The goal of this study is to analyze and describe the propagation behavior of ultrasonic waves in masonry buildings. The study consists of identifying, by unsupervised clustering methods, different wave propagation time profiles in masonry walls. The methods used are longitudinal k-means and GCKM (growth curve model combined with k-means), one being better known and widely used and the other a little more complex and with good results in the literature, respectively.