Estimation of elastic modulus of recycle aggregate concrete based on hybrid and ensemble‐hybrid approaches

Abstract The utilization of recycled aggregate concrete (RAC) within the construction sector has the potential to prevent irreversible harm to the environment and reduce the depletion of natural resources. Nonetheless, it is essential to scrutinize the quality of RAC before utilizing it in practical applications. RAC's most significant design parameter is its elastic modulus, typically determined through time‐consuming and costly experiments. Machine learning (ML) techniques can be a feasible solution to reduce the number of experiments required and obtain accurate estimates. This article employs two robust ML techniques, namely Xtreme Gradient Boosting (XGB) and Adaptive Boosting regression, in three distinct modes, namely individual, hybrid, and ensemble‐hybrid methods. Furthermore, phasor particle swarm optimization (PPSO) and chaos game optimization (CGO) have been utilized in hybrid and ensemble‐hybrid modes to optimize final results, minimize errors, and obtain highly precise outcomes. As mentioned earlier, several evaluators were employed to identify the most appropriate model to compare the modes. The findings suggest that the relevant optimizers significantly contributed to achieving superior metric values. Specifically, the results indicated that PPSO exhibited greater efficacy than CGO in optimizing outputs and enhancing accuracy. The hybrid models, particularly XGB in conjunction with PPSO, yielded the most favorable correlation coefficient and root mean square error values equal to 0.996 and 0.336 and can serve as an effective ML method to determine elastic modulus of recycle aggregate concrete for time and energy storage.