Open AccessAPPLICATION OF CONVOLUTIONAL NEURAL NETWORKS IN WALL MOISTURE IDENTIFICATION BY EIT METHOD
Author(s) -
Grzegorz Kłosowski,
Tomasz Rymarczyk
Publication year - 2022
Publication title -
informatyka, automatyka, pomiary w gospodarce i ochronie środowiska
Language(s) - English
Resource type - Journals
eISSN - 2391-6761
pISSN - 2083-0157
DOI - 10.35784/iapgos.2883
Subject(s) - convolutional neural network , artificial neural network , electrical impedance tomography , computer science , identification (biology) , artificial intelligence , moisture , deep learning , recurrent neural network , pattern recognition (psychology) , electrical impedance , engineering , geography , meteorology , botany , electrical engineering , biology
The article presents the results of research in the area of using deep neural networks to identify moisture inside the walls of buildings using electrical impedance tomography. Two deep neural networks were used to transform the input measurements into images of damp places - convolutional neural networks (CNN) and recurrent long short-term memory networks LSTM. After training both models, a comparative assessment of the results obtained thanks to them was made. The conclusions show that both models are highly utilitarian in the analyzed problem. However, slightly better results were obtained with the LSTM method.