Research On The Acoustic Characteristics And Engineering Parameter Prediction Of Deep-water Shallow Formation

Deep-water shallow formation is the undiagenetic stratum within 1000 m below the mudline.The shallow formation has the characteristics of soft soil,high water content and low strength.Its engineering strength has an important impact on the wellhead stability,the depth of the conductor into the mud,the laying of the submarine pipeline,the calculation of the foundation mooring force,and the design of the central manifold.Conventional gravity coring and borehole coring have high cost,few sampling points,long period,and will cause soil disturbance.Therefore,it is of great significance to accurately predict shallow geotechnical engineering parameters before drilling by using acoustic seismic.In order to study the influence of deep-water and shallow geotechnical engineering parameters on acoustic characteristics,calculation and analysis were firstly carried out at the theoretical level.Due to the complexity of the wave equation,it is impossible to directly establish the mathematical relationship between the acoustic wave characteristics and the four geotechnical parameters.Based on Biot’s acoustic wave propagation theory and Berryman’s self-consistent theory,the coefficients in the wave equation were used as bridges to establish the relationship between acoustic characteristics and engineering parameters.The acoustic wave velocity and acoustic attenuation under different geotechnical engineering parameters were calculated using Matlab software.The optimal theoretical relationship between acoustic characteristics and geotechnical engineering parameters was determined according to the principle of mathematical statistical analysis.A small-scale indoor simulated acoustic experiment based on the similarity principle was implemented,the shear strength,acoustic velocity,and acoustic attenuation of soil layers with different clay content,density,water content were tested.Through the simulation results,the theoretical calculation results of acoustic characteristics were verified and revised based on field data,the single-factor formula of acoustic characteristics and shallow geotechnical properties were established.The influence weights of different geotechnical parameters on acoustic velocity and acoustic attenuation coefficient are studied by variance analysis method.It is found that the four deep-water shallow formation engineering parameters studied in this paper have non-negligible effects on acoustic characteristics.Combined with the single factor formula,the multiple linear regression method was transformed,a multi-factor relationship model was established among the acoustic properties(acoustic velocity and attenuation)and shallow formation properties(sand-viscosity ratio,water content,density and shear strength).The database were established by combining the theoretical calculation results,experimental results and field measured data.Taking shear strength as an example,the Tensorflow machine learning framework was used to build Support Vector Regression machine,fully connected Deep Neural Network,Random Forest,Light GBM and XGBoost machine learning models.These models can be used to predict the engineering parameters of deep-water shallow soils from acoustic features.The evaluation indicators of five machine learning models were compared,and it was found that the Random Forest model has the best prediction effect on the shear strength.A method for predicting the geotechnical properties of shallow formation in deep-water by using acoustic characteristics was formed and compiled into software.Through field case analysis,it is found that the prediction accuracy of this method reaches 83%.This study provides a new way for predicting geotechnical parameters of shallow formation before drilling.