基于BP神经网络的深部复合岩体变形模式预测

打开文本图片集

中图分类号：TU458 文献标志码：A 文章编号：2095-2945（2025）18-0057-05

Abstract：Predictingthedeformationpaternsof tunnelsurroundingrockisofsignificanttheoreticalandpracticalimportance forguidigtunnelconstruction，dynamicprotection，andensuringthelong-termsafeoperationoftunnels.Topredictandclssify thedefomationpaternsof depcompositerockmases，thisstudyutilizesdatafromcompositerock massmodeltestsandDIC （DigitalImageCorelation）deformationanalysis.ABP（BackPropagation）neuralnetworkisemployedtoestablishthecomplex nonlinearrelationshipbetweentheglobaldeformationofthesuroundingrockandthedeformationpattems，therebyachieving intelligent prediction of the deformation modes.The results show that the coefficient of determination （R2 ）between the predicted valuesandtheactualvaluesisO.996，indicatingahighpredictionaccuracy.Thetrainedneuralnetworkdemonstratesastrong capabilityinpredictingthedeformationpattrnsofthesuroundingrock.Theresearchresultscanprovideanefectivemethodfor predicting deformation and fracture in practical engineering.

Keywords：neuralnetwork;compositerockmas;deformationandfracturemode;DICdeformationandanalysisdata; modelraining

受地质环境、岩体强度和高地应力的综合作用，当深埋隧（巷）道穿越软弱破碎岩层时极易发生围岩挤压大变形，常规支护措施难以有效控制，严重时还会导致支护损毁乃至坍塌等后果，对人员安全与货物运输造成威胁]。（剩余5760字）