Brick wall moisture evaluation in historic buildings using neural networks

The article presents the results of numerical analyses and experimental research concerning the neural evaluation of the mass moisture content U-mc of brick walls in historic buildings. For the purpose of training, testing and validating artificial neural networks, a representative data set was built on the basis of tests of the moisture content and salinity of brick walls in ten historic buildings. The article presents two structures of artificial neural networks that are most useful for the neural evaluation of the mass moisture content, which were selected on the basis of the conducted analyzes. The results of comparative applications of all analysed algorithms were also included in the paper. High R-2 values for learning, testing and validation using artificial neural networks prove the credibility of the results. This means that the proposed method can be used in construction practice to assess, after practical verification, the moisture content of brick walls.

Automated summary

The article presents the results of numerical analyses and experimental research on the neural evaluation of the mass moisture content of brick walls in historic buildings, demonstrating that the proposed method can be used in construction practice to assess the moisture content of brick walls.