Integrating multiscale numerical simulations with machine learning to predict the strain sensing efficiency of nano-engineered smart cementitious composites

Prediction of in-situ strain sensing efficiency self-sensing cementitious composites using machine learning (ML) requires a large, representative, consistent, and accurate dataset. However, such large experimental dataset is not readily available. Moreover, the success ML approach depends on its ability to abide by fundamental laws physics. To address these challenges this paper synergistically integrates validated finite element analysis (FEA)-based multiscale simulation framework with predict strain-sensing enabled incorporating nano-engineered interfaces. The leveraged develop balanced, complete, consistent containing 3000 combinations strain-dependent electromechanical responses. This used nanoengineered feed-forward multilayer perceptron-based neural network (NN) which shows excellent prediction efficacy. also applies Shapley Additive Explanations (SHAP) algorithm interpret NN predictions in light relative importance different design parameters composite. Overall, synergistic comprehensive presented here can be as starting point toward development reliable performance standards accelerate acceptance for large-scale applications.