A Non-Parametric modelling of Concrete Mixtures for Optimum productivity

Abstract

The absence of the distribution assumption in the modeling of any physical quantity enhance the quality of modeling. The complexity in concrete mixture modeling becomes critical in optimum design and control of concrete mixture. In this paper, a nonparametric modeling technique is adopted to avoid the guessing assumption noted with parametric. In order to address this condition, Kernel estimator techniques is considered as a result of its efficiency to model such a task as concrete mixture distribution. A thousand and thirty (1030) portion mixes for each of the two segments was considered, particularly the water to bond proportion (w/c) with a range (0.27–1.88), and the fine concrete quantities with a scope of (140–540), estimated in kilograms(Kg) were embraced. The performance of the kernel estimator is measured by using the Integrated Square Errors (ISE) for the various smoothing techniques. The paper considered seven kernel estimators and the most performed estimator is evaluated based on the ISE testing technique. By means of observation, the Epanechnikov kernel estimator was considered the most appropriate with ISE of about 7.33%, thus allowing the water to bond proportion to have a major effect on the concrete mixture.