| With the fast growth of economy and technology,the development and utilization of urban aboveground space has been nearly saturated.It is urgent to carry out the investigation,evaluation and zoning of urban underground space resources.How to investigate and evaluate the urban underground space resources,the development and utilization of underground space and the environmental geological problems that may arise need to be further studied.Under this background,how to obtain the engineering geotechnical index of underground space efficiently and economically is an important topic of underground space evaluation.Xixian new area is located in the Quaternary strata,the sedimentary environment is more complex,geological,engineering geological and other factors restrict the development and utilization of underground space.On the basis of systematically combing the modeling theory and methods,this paper uses Depth Insight 3D geological modeling software to establish the deterministic 3D geological structure model and the stochastic 3D attribute model of lithology and logging in Xixian new area;Based on the correlation between logging parameters and collapsibility coefficient,a neural network model is constructed to estimate the collapsibility coefficient by 7 logging parameters,and the collapsibility zone of loess layer in the late Pleistocene in Xixian new area is defined.The main results are as follows:(1)A three-dimensional deterministic geological structure model is established.Based on multi-source data such as borehole data,geological map and DEM,a three-dimensional geological structure model of nine Quaternary strata in Xixian new area is established by using depthinsight modeling software.Three sections are selected to test the results.The results show that the strata and fault extension of the model are basically consistent with those of the section,and the model results are true and reliable Fence map and other three-dimensional visualization display.(2)Based on the interpolation method of sequential simulation,the lithology stochastic model and 11 logging parameter stochastic models are established respectively.In the framework of geological structure model,according to the engineering geological classification,four basic types of gravel soil,sandy soil,cohesive soil and loess are established,and the results are tested by means of section comparison.Similarly,11 random models of logging parameters are established after processing the total factor logging data in the study area.The modeling results are visualized,and the performance characteristics of logging parameters have a strong correlation with the formation lithology,which changes with the change of lithology.(3)The artificial neural network method is used to estimate the collapsibility coefficient based on multi-source logging data.The distribution characteristics of 11 logging parameters and collapsibility coefficient data are analyzed and standardized.The correlation between logging parameters and collapsibility coefficient is studied by statistical method.The collapsibility coefficient of late Pleistocene loess layer is estimated by using 7 logging parameters with high correlation and multi-layer perceptron model.The results show that the average proportional error is 12.8%.(4)The data body of logging attribute model in the loess layer of late Pleistocene was extracted by MATLAB program.The results of neural network simulation were used to obtain the data body of collapsibility coefficient in loess area,and the average value of collapsibility coefficient was calculated to divide the collapsibility zone.The area of the non collapsible loess layer in the late Pleistocene in Xixian new area is 56.83km~2,accounting for 7.46%,mainly distributed on Xianyang plateau;The area of slight depression is 704.22 km~2,accounting for92.39%,and both Xianyang plateau and alluvial proluvial plain are distributed;The area of medium depression is 1.11km~2,accounting for 0.14%,mainly distributed in alluvial and proluvial plain. |
PDF Full Text Request