Abstract
To produce highly effective electrically conductive cementitious composites and advance non-destructive structural-health monitoring techniques, graphite-based nanomaterials (GrNs) are viable conductive fillers. A comprehensive database was created for CS and ER pertinent to GrNCC’s experimental studies from open sources. It is a challenging multi objective optimization problem (MOOP) due to the requirement of both mechanical strength and electrical resistivity for the design of Graphene nanoparticles-reinforced cementitious composites (GrNCC). The current study provides an ample data-driven approach employing machine learning (ML) techniques and a non-dominated sorting genetic algorithm (NSGA-II) to address this multi objective design optimization (MODO) problem for GrNCC. ML theories quantify how important parameters affect GrNCC’s qualities. First, using processed experimental datasets, gene expression programming (GEP) is used to develop prediction models for GrNCC’s uniaxial compressive strength (CS) and electrical resistivity (ER). The findings demonstrate its superior predictive accuracy, as evidenced by high R2 of 0.92 and 0.91 and low mean absolute error (MAE) scores of 1.57 MPa, and 60.46 kΩ.cm for CS and ER, respectively. For the MODO technique, parameter analysis was performed, which aids in identifying the parameters that need to be optimized and defining their constraints. Finally, NSGA-II is used as the foundation for the MODO technique. With the proposed prediction models as objective functions, it simultaneously optimizes the CS and ER of the GrNCC. It successfully produces a collection of Pareto solutions that make it easier to choose the right parameters for the GrNCC design. The proposed MODO technique can be efficiently applied to multiobjective design and optimization of electrical and mechanical properties of GrNCC.
[Display omitted]