Pattern Recognition and Oxidation Classification in Metal Structures of Industrial Roofs Using Artificial Intelligence

The overall cost of structural corrosion is estimated to be in the order trillions dollars throughout world. This a fault that can bring numerous harms industries through water infiltrations, roof breaking and collapsing structures, even create work-related accidents. In this research, an artificial intelligence was developed evaluate aerial images factory roofs, aiming towards more accurate efficient maintenance. Previous studies were conducted optimal flight conditions unmanned vehicle such as ambient illuminance, cruise speed height. Afterwards Taguchi analysis made. Based on knowledge building maintenance, neural network elaborated containing three great levels its architecture. first step, extractor model applied obtain only information from roofs captured images. Next, second stage included detection segmentation. third stage, performed criticality classification detected failures. proposed system trained evaluated total 249,000 m2 constructed area results presented potential application for diagnosis corrosions metal structures.