Smart prediction of ground displacement using parallel neural network models and high-precision geodetic measurements

The goal of this research was to investigate the possibilities of using parallel neural network models in combination with high-precision geodetic measurements to predict ground displacement. The relevance of this task is defined by the growing need for effective methods to ensure safety and reduce risks associated with the operation of engineering facilities. A methodology for smart prediction has been developed within the framework of the research, based on the analysis of a complex data set using parallel neural network models. Results of geodetic monitoring based on selected 120 observation points during the period of 2 years formed the empirical basis. The key results include an improvement in the accuracy of ground displacement predictions up to 95% (SE = 1.2; p < 0.01); a reduction of the data processing time by 3.5 times (t = –14.8; p < 0.001); a reduction of the number of errors by 28% (F = 23.4; p < 0.01). The practical value of the developed approach is associated with the possibility of its application to minimize risks and costs in the management of complex infrastructure facilities. The theoretical significance consists in the development of methodology for real-time smart analysis of Big Data to predict dynamic processes.