An extended multi-model regression approach for compressive strength prediction and optimization of a concrete mixture

Due to the significant delay and cost associated with experimental tests, a model-based evaluation of concrete compressive strength is high value, both for purpose prediction as well mixture optimization. In contrast prior recent studies employing single regression model, in this paper, we present combined multi-model framework where methods based on artificial neural network, random forest polynomial are jointly implemented higher accuracy. The outcomes individual models via linear weighting strategy optimized over training data set quadratic convex optimization problem. It worth mentioning that due convexity formulated problem, globally optimum obtained standard numerical solvers. Afterwards, multi-objective genetic algorithm-based method proposed under practical constraints, Pareto front cost-CS trade-off has been available set. Numerical evaluations show achieves significantly accuracy, i.e., approximately 18% reduction mean squared error, without weight optimization, roughly 30% error an combination following compared best model multi-layered network.