Orthogonal Experiments and Neural Networks Analysis of Concrete Performance

Abstract

In order to explore the possibility that adding an appropriate amount of alkaline activator into fly ash cement may improve the early activity of fly ash and ensure the strength performance of concrete, this study analyzed the influence of 0–30% fly ash substitute on the early and late (3–28 days) compressive strength of concrete by using three methods, namely, the concrete laboratory test, orthogonal test, and neural network, under the condition of 0.5 water binder. We obtained the following results: (1) The strength of the concrete mixed with fly ash at the same alkali and the same age decreases with the increase of fly ash content and decreases with the decrease of age; the strength is the highest when the alkali content is 6% or 5%. (2) The higher the content of fly ash, the lower the strength of the mixture, and the greater the decrease of the early strength of the mixture, while the optimum dosage of NaOH is the same. (3) Orthogonal experimental design can be effectively used to analyze the primary and secondary degree of each factor and the best combination of them (cement, fly ash, NaOH, standard, water, etc.). (4) High correlations between the compressive strength and the component composition of concrete can be obtained using the prediction abilities of the neural networks. The above test results show that on the basis of the concrete compressive strength test, the comprehensive application of the orthogonal test and the neural network method can be used to analyze the relationship between strength and the variables and to test the influence of the variables and their interaction on concrete strength, and the results are accurate and reliable.