Study On Strength And Dry Shrinkage Model Of Fly Ash Concrete Based On Neural Network

Compressive strength is an important index of concrete quality,and dry shrinkage will cause concrete cracking,seriously lead to accidents.Therefore,compressive strength and dry shrinkage play an important role in the study of concrete performance.At present,the commonly used concrete estimation methods of strength and shrinkage are mostly traditional empirical formulas.Whether it is suitable for fly ash concrete is still to be verified,the accuracy and reliability fluctuate greatly as well.According to the experimental data in some scholars,a prediction model of strength and shrinkage performance of fly ash concrete are established based on neural network in this paper.And the development law of strength and dry shrinkage of fly ash concrete has been studied by the prediction model.The main conclusions are as follows.1.The training,test and validation test samples show a powerful ability in learning and generalization of the strength and dry shrinkage prediction model,and the calculation accuracy reached 5%and 10%,The experimental results show that the prediction results of neural network model are reliable and the accuracy is high.The fitting effect of neural network model is better than that of domestic and foreign commonly used intensity and shrinkage models.2.The strength model shows that the strength of fly ash concrete is low at early stage that the strength at 3 d is only about 22%of the design;followed by the rapid growth,the strength at 7 d meet the design strength of 72%;the late strength is increasing,but gradually stabilized after 90 d,the 90 d and 540 d of the maximum intensity difference was only 8.1 MPa.3.The dry shrinkage model shows that the early drying shrinkage of fly ash concrete is small.It is only about 2.7%of total shrinkage among the first year at 3 d,but it occurs rapidly.It is close to the half at 28 d and nearly 90%at 90 d.In general,the dry shrinkage of fly ash concrete is a dramatic change before 90 d,and then it is slow.4.Based on the BP neural network model,the strength values of different cement grade,water cement ratio and fly ash replacement rate are deduced.The results show that the higher cement grade and the lower water cement ratio lead to a high strength of fly ash concrete.The strength is improved while the replacement of cement in the range of 0 to 5%.The strength of the concrete decreases with the increase of the content when the content of fly ash is more than 5%.When the fly ash replaces fine aggregate,the replacement rate is proportional to the strength of fly ash concrete.5.Based on the BP neural network model,the dry shrinkage values of water binder ratio,sand ratio and fly ash replacement rate are deduced.The results show that the lower water binder ratio and sand ratio and the larger fly ash content lead to the small dry shrinkage of fly ash concrete,but the influence of sand rate is weak.