Study On Strength Prediction Model Of Clay With Different Plasticity Exponents Under Thixotropy

With the development of China’s economy and the growth of population,the development strategy of many cities is to expand from outside to cut mountains and fill ditches to build cities.Therefore,many filling projects are encountered in infrastructure construction.Fill engineering problem is considered in the strength of the filled soil,so under the background of this project,consider the long-term strength of the fill soil,also is the problem the thixotropy of the loess body,as the growth of the time,it has certain growth and soil strength tends to be stable,only change rate stable and even the same intensity,Only in the filling area can the building foundation stably bear the building load.Therefore,starting from clay with different plasticity index,this paper studies the strength prediction model of clay with different plasticity index under the influence of thixotropy,so as to provide theoretical guidance for the construction of infrastructure in the later stage of filling project.In this paper,the compressive strength test and direct shear test of artificial clay under the influence of thixotropy are carried out.From the aspects of plasticity index,age,dry density and water content,the influence on the strength of clay is considered,and the multiple regression prediction model and the neural network prediction model are established by using the mathematical analysis method.The main conclusions are as follows:(1)Based on different plasticity index of artificial soil compaction test resultsshow that with the increase of plasticity index,maximum dry density decreases,the optimum moisture content increases gradually,because of themontmorillonite mineral powder has a larger specific surface area,enhanced the new soil water imbibition,lead to the optimal water content increases,after combined with sandy loam,form the skeleton structure,increasing the porosity of the soil,Therefore,the maximum dry density is reduced.(2)The results of unconfined compressive strength test and direct shear test on artificial clay show that the range of unconfined compressive strength and shear strength(cohesion)increases gradually with the increase of plasticity index.Then,the proportion of each factor is calculated by using the ratio of sensitivity to ratio of each factor.The results show that the weight of each factor for clay strength is also different,in which the weight of dry density is larger and the weight of age is smaller.(3)Under the basis of the principle of least square method,the test results of the multivariate nonlinear regression model is established,and then analysis the different between the plasticity index,age,dry density,moisture content of the single factor effect on the unconfined compressive strength,the influence of the study found that as the growth of the plasticity index,age,dry density,compressive strength increases gradually,effect was positively effect,However,the effect of water content is negative.With the increase of water content,the unconfined compressive strength decreases gradually.Moreover,with the increase of age,the compressive strength increases rapidly before 20 days,and then grows slowly and becomes stable after 20 days.(4)From the effects of various influence factors of shear strength of regression analysis found that the dry density,age has promoting effect on the growth of the cohesion of the moisture content reduces the growth of cohesion,cohesion of the plasticity index of slower growth,between 8.8 and 13.1 after the plasticity index greater than 13.1 is growing rapidly,and plasticity index to the cohesion of influence,It fully shows the contribution of clay content to the strength of soil sample,and the selection of soil material with appropriate plasticity index in filling engineering will greatly enhance the strength growth of soil and the formation of soil structure.(5)On the basis of the principle of artificial neural network,USES the error propagation algorithm on the compressive strength of clay before BP neural network model is established and its influence factors,through the middle layer neuron nodes of BP neural network to determine,as well as the learning of neural network training and testing,the accuracy of neural network to achieve the required,And the compressive strength of clay can be well predicted.However,compared with the multivariate nonlinear regression model,the regression model can more intuitively reflect the influence of various factors and coupling effects on the unconfined compressive strength.Moreover,by comparing the R2 value of the two models,the determination coefficient of the regression model is slightly higher than that of the BP neural network,so the accuracy of the regression model is higher.(6)The shear strength of clay is also analyzed and predicted by two prediction models.The determination coefficient of the two models is greater than 0.9 and the difference is not significant,indicating that the accuracy of the two models is basically the same.The prediction values of 8 groups of experimental values and two models,as well as their relative errors,were analyzed and compared to verify the accuracy of the two models.It was found that compared with the regression model,the relative errors of the experimental values and predicted values were smaller,and the BP neural network model was more accurate.