Experimental Study On Early Estimating The Strength Of Fly Ash Concrete

With the development of the concrete technology, a lot of mineral additive and admixture are added in the concrete, especially the large-scale utility of fly ash become the focus.The early quality control of fly ash concrete becomes more and more important.Whether the conventional early estimating method of concrete strength are applicable for the modern concrete and a new better method and forecast model can be founded to estimate the strength of the concrete, are both drawing extensive attention. In this circumstance, the paper makes completely research on early estimating of fly ash concrete strength, and puts forward a new way to forecast the strength of concrete using the neural networks. The paper supplies reference for estimating the concrete strength quickly and controlling the quality of the concrete which is filled with fly ash and super plasticizer.First, this paper makes general fly ash concrete, high-strength-fly ash concrete, high-fluidity-fly ash concrete and high-scale-fly ash concrete, and gets the data of boiling strength and 28d strength, then sets up three math models based on the experiment data, which apply simple regression analysis, bivariate regression analysis, nonlinear regression analysis separately by the MATLAB. The theoretical analysis to the accelerated curing experiment provides gist for early predict of fly ash concrete strength.Second, the paper expounds the theory of neural networks and the program, and sets up a new model by BP neural networks based on the experiment data, validates the precision by the experiment data, puts forward the new way to predict the concrete strength early using the BP networks combined with the experiment. Meanwhile, the paper establishes the GRNN networks. Comparing the BP networks with the GRNN networks, the result is that two networks are fit for predicting the concrete strength early, the BP networks has better precision, but, the GRNN networks has better stability.