A novel approach for U-value estimation of buildings' multi-layer walls using infrared thermography and artificial intelligence

Estimating the U-value of walls of buildings is a key process to evaluate the overall thermal performance. Low U-value in buildings is desired in order to keep heat within the envelop and consume less energy in heating. Addressing the limitations in the currently used U-value estimation techniques, this paper proposes a novel approach for estimating the U-value of the envelop of buildings using infrared thermography and Artificial Neural Network (ANN) with the application of a point heat source. The novel system is calibrated by training the ANN in a lab environment using a wide range of samples with multi-layers to be able to estimate the in-situ U-value of walls in real buildings during field work with relatively high accuracy.