Prediction Of The Calcium Hydroxide Content Of Cement-based Composites Based On BP Neural Network

Fly ash and slag are the most widely used two kinds of mineral admixtures now,they can significantly improve the performance of concrete,save cement and reduce the engineering cost,it has good economic benefits.With the continuous development of concrete technology,the content of mineral admixtures in concrete increases gradually,but if the content is too big,it can lead to calcium hydroxide insufficient in concrete,seriously affect the carbonation resistance of concrete.Therefore,it has very important significance for assessing the durability of concrete structure to accurately predict calcium hydroxide content of mineral admixtures in concrete.Based on above situation,this paper mainly expounds the prediction problem of 28 d calcium hydroxide content of the cement-fly ash cementing material and the cement-ground slag cementing material.Two kinds of cement-based composites are considered for four factors: the content of fly ash(the content of ground slag),fly ash type(ground slag type),cement type and water-binder ratio.Among them,the content of fly ash(the content of ground slag)is 0%,30 % and 60%,water-binder ratio is 0.36,0.43 and 0.50.The test scheme was determined according to the orthogonal design,calcium hydroxide content was determined by DSC-TG and the influence degree of each experimental factor to the calcium hydroxide content was determined by range analysis and variance analysis,then selected suitable raw materials and proportioning parameters on this basis.The functional relationship was established between parameters and calcium hydroxide content by BP neural network,applied test datas to prediction function and compared the predicted values and the experimental measurements.The results show that it can use some parameters to predict the calcium hydroxide content of the cement-fly ash cementing material,the parameters are water-binder ratio,CaO content of cement,SiO2 content of cement,Al2O3 content of cement,Fe2O3 content of cement,specific surface area of cement,strength of cement,the content of fly ash,CaO content of fly ash,SiO2 content of fly ash,Al2O3 content of fly ash,Fe2O3 content of fly ash,LOI of ash fly and specific surface area of fly ash;It can use other parameters to predict the cement-ground slag cementing material,the parameters are water-binder ratio,CaO content of cement,SiO2 content of cement,Al2O3 content of cement,Fe2O3 content of cement,specific surface area of cement,strength of cement,the content of ground slag,CaO content of ground slag,SiO2 content of ground slag,Al2O3 content of ground slag,MgO content of ground slag,LOI of ground slag and specific surface area of ground slag.The prediction errors of two functions can be controlled within 10%,this method would provide a new idea for the prediction of calcium hydroxide content of cement-based composites.