Physical cellular automata and artificial fish swarm fusion catastrophic failure prediction of brittle-like materials

Abstract

The investigation aims to establish a physics-informed data-driven method for catastrophic failure analysis of brittle-like materials based on the strategy of the physical cellular automata and artificial fish swarm fusion prediction. Based on the method, local mesoscopic elasticity modulus and damage distribution of brittle-like materials can be predicted based on the monitored data of the global macroscopic mechanical response. In the data-driven prediction process based on the simplification of the artificial fish swarm algorithm, physical catastrophic failure mechanisms can be considered by utilizing a modified cellular automata technique for updating and iterating damage distribution based on the principle of energy conservation and energy dissipation. Two numerical cases are implemented to support the developed method. The results support that the method possesses a good convergence and a high prediction accuracy. The maximum prediction error of the macroscopic stress-strain relationship of the two numerical cases is 9.6% based on the optimal mesoscopic elasticity modulus prediction. Meanwhile, the predicted local mesoscopic catastrophic failure paths of the two numerical cases both match well with the corresponding experimental results. The developed physics-informed data-driven method can provide an efficient tool to predict both macroscopic and mesoscopic failure particularities of brittle-like materials for better investigating their catastrophic failure mechanisms.