Experimental Study On Solidification Strength Of Dalian Marine Soft Soil And Its Neural Network Prediction

With the advancement of urbanization in my country and the increasing scale of construction in coastal areas,the treatment of Marine soft soil has become a problem that must be faced in the process of engineering construction.Marine soft soil is characterized by high water content,large compressibility and low strength,so it cannot be directly applied to engineering construction due to its poor engineering properties.In addition,the scale of my country’s infrastructure construction projects is huge,and the corresponding amount of Marine soft soil is also very large.If it is not treated,there will be serious problems such as occupying a lot of land,producing silt foundation and even affecting the marine ecological environment.Therefore,how to make good use of Marine soft soil reasonably has important engineering value and practical significance,and conforms to the concept of sustainable development in my country.At present,it is an effective method to improve the properties of soft soil by adding inorganic curing agent and industrial wastes.Cement is relatively abundant,cheap and efficient,and is a good curing agent material;As an additive,silica fume can reduce the dosage of cement,improve the performance of cement soil,and contribute to industrial waste recovery;Calcium hydroxide,as a common alkaline activator,can also be used in the study of soft soil solidification.In this paper,the solidification strength of Marine soft soil in an engineering site in Dalian was studied with cement,silica fume and calcium hydroxide as curing agents.Firstly,the solidification mechanism of soft soil strengthened by curing agent was briefly described.Then,the stress-strain relationship curve of solidified soft soil was obtained through unconfined compressive strength test.The influence of curing agent content,water content and curing age on the unconfined compressive strength of solidified soft soil was studied.At the same time,an orthogonal test was designed to analyze the effect of curing agent on the strength index of solidified soft soil and the optimal ratio of curing agent by means of statistical methods such as range and variance.Furthermore,the artificial neural network technology was applied to the test results,and the nonlinear relationship between the content of curing agent(cement,silica fume,calcium hydroxide),moisture content,curing age and the unconfined compressive strength of solidified soft soil was established through MATLAB software.BP and RBF neural network models for strength prediction of solidified soft soil are trained by using orthogonal test data.It can predict the strength of Marine soft soil after solidification to a certain extent,replace part of the test,reduce the test workload,save manpower and material resources,and facilitate engineers to control the overall quality.At the same time,the prediction model of cement content based on BP neural network is established,which improves the design efficiency of soft soil curing agent,and provides a new way for the study of the ratio of soft soil curing agent.