FULL CONNECTED NEURAL-NETWORK FOR SIMULATION OF EXTANTION IN SELF-STRESSED MONOLITIC SLABS ON GROUND

Abstract

In this article the strategy of interdisciplinary convergence of mechanics and artificial intelligence is illustrated. The article presents the results of calculating displacements in self-stressed monolithic slabs on ground obtained using a trained fully connected neural network. The empirical results of displacements in slabs on ground, displacements calculated according to the physicomechanical model, and obtained using a neural network are represented. The inspiration brought us to study neural networks modeling biological neural networks are follow: neural networks can autonomously detect patterns hidden in phenomena and can identify parameters on complex behavioral tracks of different physical systems. The authors describe in detail the developed and trained fully connected neural network.