Automatic analysis of GPR scans on concrete structures

摘要

Current Ground Penetrating Radar (GPR) systems can produce great quantities of data when used to scan concrete structures. This data must be interpreted by an experienced operator before it can be useful to a practising engineer. This can be both a laborious and time consuming task. Ongoing work in the UK at Liverpool University aims to develop an automatic system which uses artificial neural network technology to identify the significant features in a radar scan and deduce the nature of the objects that caused them. The principle aims are to identify and evaluate the depth to reinforcing bars and voids, as well as their sizes, and to characterise voids as air filled or water filled.