BULETINUL INSTITUTULUI POLITEHNIC DIN IAŞI Publicat de Universitatea Tehnică „Gheorghe Asachi” din Iaşi Volumul 62 (66), Numărul 2, 2016 Secţia CONSTRUCŢII. ARHITECTURĂ USING THERMOGRAPHY AND ARTIFICIAL NEURAL NETWORKS IN CIVIL ENGINEERING BY ALEXANDRINA-ELENA PANDELEA * and MIHAI BUDESCU “Gheorghe Asachi” Technical University of Iaşi, Faculty of Civil Engineering and Building Services, Received: June 12, 2016 Accepted for publication: July 27, 2016 Abstract. By applying the thermography and neural networks, a diagnosis of the heat loss can establish, followed by the state of a building. The infrared images recorded by the camera, obtained through the thermography, will be the input data for the artificial neural network. For this type of problem, a feed- forward, multilayer, supervised neural network is adopted and trained with a back-propagation algorithm. The activation function used in this matter is a function specific to the information classification problems, namely the step function. Keywords: diagnosis; heat loss; back-propagation algorithm; surface; temperature. 1. Introduction Thermal methods involve a temperature variation analysis of building elements used to detect degradations and faults of the building elements, material particularity irregularities, detecting thermal bridges etc. In civil engineering, thermography is considered an effective technique for non-destructive tests, without contact on the structure and is used to assess the performance of the building envelope. Any civil building can be evaluated in a very short time and the aim of this evaluation is to check if there are any * Corresponding author: e-mail: pandelea.alexandrina@gmail.com