基于神经网络的隔湖组黏性土工程地质数据分析

打开文本图片集

中图分类号：P642.2 文献标志码：A 文章编号：2095-1329（2025）02-0080-07

Abstract： Cohesivesoil inGehuformatioisthefoundationbaringlayermostusedintheSuzhou-Wuxi-Changzouarea，anditsalso oneofthe mostcommonlyusedstratuminundergroundspacedevelopment.Weselected126geologicalboreholes toresearchthisstratum using 446 groupsofsampledata （400 groups oftrainingdataand46groups ofpredictiondata）.Weuseda neuralnetworkalgorithmto establisharelationshipmodelbetweenphysicalparametersand mechanicalparameters.Wetheninput46groupsof predictiondatainto themodeltoverifyitsreliabilityandalsousedspecificdatatoresearchtherelationshipbetweenphysicalandmechanicalparameters. Theresultsshowtat：（l）TeearsoncoelationcoefcientbetweenthepredictionrsultsandmeasureddataisoretanO8bter thantheftingesultsofmultiplelinearegressionemodelisverifedaselable.（2）Tereiscertaincoelationbetwenpyical andmechanicalparameters，andtheneuraletworkalgoritmcanexpresscoelationmodelsbetweenmultipleparameters.eneral networkalgorithallows teuderstandngofthehighvalueinteutilizationofengenggeologicalbigdata.Temethodissiple and reliable，and it can be applied in the research and analysis of data.

ey words：engineering geological; cohesive soil; Gehu formation; neural network; data analysis

对其进行更深层次的挖掘。（剩余7726字）