Abstract: The article is devoted to the development of an algorithm for early detection of defects at construction sites, which is
used in geotechnical monitoring systems to improve the accuracy of predictive estimates of the stability of structures. The emerging
design situations that lead to accidents and catastrophes of a man-made and natural nature show the need for further development of
algorithmic support for geotechnical monitoring systems, despite the existing developments in the field of geotechnical monitoring
and assessment of the stability of geotechnical systems. A flowchart of an algorithm for early detection of defects at construction
sites is presented, based on the author's approach to identifying key points of geotechnical monitoring, methods of bifurcation theory,
as well as neural network analysis. A distinctive feature of the developed algorithm, in addition to using a neural network to adjust to
geotechnical features, is the possibility of dynamically adjusting the ranges of variation of the stability limits of the geotechnical
system laid down in the design documentation for the construction object under study. The results of the practical application of the
developed algorithm in the geotechnical monitoring system (observations were carried out from 2016 to 2021) of the parameters of
the soil base, as well as the physical and mechanical parameters of the structural elements of the foundation and structure are
described. The structure was a three-storey building erected on a brick ribbon foundation. The object of research is located in the city
of Murom, Vladimir region. During the application of the developed algorithm, estimates of the places of defect formation at control
points and their localization were obtained, which were confirmed during further observations. The developed algorithm can be used
in geotechnical monitoring systems throughout the entire life cycle of a geotechnical system.
Index terms: algorithm, control, process, geotechnical system, destructive.