Integrating machine learning with multi-objective optimization to maximize the ductility of macro-synthetic fiber reinforced rubber concrete

The brittleness of concrete is mitigated through the incorporation of waste rubber particles (RP) and macro-synthetic fibers (MF), and an intelligent framework is proposed to guide their dosages. The modification mechanisms of RP and MF on macro-synthetic fiber reinforced rubber concrete (MFR-RuC) are investigated through experimental analysis based on the energy evolution law. An intelligent framework, integrating the kernel extreme learning machine (KELM) with the non-dominated sorting genetic algorithm II (NSGA-II), minimizes the proportion of elastic strain energy ( K ) and maximizes the ultimate compressive strain ( ɛ u ) and total absorbed and dissipated energy ( U tot ), with the prerequisite that the compressive strength ( f c ) of MFR-RuC is ensured to be no less than 20 MPa. The results demonstrate that RP moderates and slows energy dissipation, and MF extends the energy dissipation period by enabling the concrete to continue bearing loads after cracking. The use of either RP or MF leads to a noticeable reduction in the brittleness, with a synergistic effect observed when both are combined. Furthermore, the implicit relationship between the performance of MFR-RuC and the dosages of RP and MF, modeled using KELM, is both accurate and reliable. Finally, the optimal usage of RP (21.01 %) and MF (0.81 %) for maximizing ductility where ɛ u , U tot , and K improved by 90.22 %, 6.61 %, and 39.07 %, compared to the control group (R0F0) are achieved using the KELM-NSGA-II intelligent framework and the TOPSIS method. A novel and sustainable approach is provided to mitigate the brittleness of concrete, with the proposed framework facilitating the intelligent development of building materials.