文章编号：1007-2993（2023）04-0392-05基坑变形在数据滤波处理条件下的组合预测研究韩国锋（陕西能源职业技术学院，陕西咸阳712000）【摘要】为合理评价基坑施工过程的变形规律，以其变形监测成果为基础，先通过双树复小波实现其变形数据的滤波处理，再通过GWO-RVM模型、Arima模型及混沌理论实现其分项组合预测。实例分析表明，双树复小波可将基坑变形数据有效分解为趋势项分量和误差项分量，并通过模型参数优化处理，可进一步提高分解效果，较传统小波具有更强的滤波能力；同时，各类分项预测模型在不同变形分量中的适用性也较强，所得组合预测结果的平均相对误差在2%左右，明显优于传统预测模型，验证了组合预测思路在基坑变形预测中的适用性，为基坑变形发展规律研究提供了一种新的思路。【关键词】基坑变形；滤波处理；混沌理论；变形预测；变形规律【中图分类号】TU196【文献标识码】Adoi:10.3969/j.issn.1007-2993.2023.04.003CombinedPredictionofFoundationPitDeformationundertheConditionofDataFilteringHanGuofeng(ShaanxiEnergyInstitute,Xianyang712000,Shaanxi,China)【Abstract】Inordertoevaluatethedeformationlawoffoundationpitconstructionprocessreasonably,basedonthedeformationmonitoringresults,thefilteringofdeformationdatawasrealizedbydoubletreecomplexwavelet,andthenthesub-combinationpredic-tionwasrealizedbyGWORVMmodel,ARIMAmodelandchaostheory.Theanalysisoftheexampleshowsthatthedeformationdataoffoundationpitcanbeeffectivelydecomposedintotrendanderrorcomponentsbydoubletreecomplexwavelet.Thedecompositioneffectcanbefurtherimprovedbyoptimizingthemodelparameters,whichismorepowerfulthantraditionalwavelet.Atthesametime,theapplicabilityofvarioussubitempredictionmodelsindifferentdeformationcomponentsisalsostrong.Theaveragerelativeerrorofthecombinedpredictionresultsisabout2%,whichisobviouslybetterthanthetraditionalpredictionmodel.Itverifiestheapplicabilityofthecombinedpredictionmethodinthedeformationpredictionoffoundationpit,andprovidesanewwayfortheresearchofdeform-ationdevelopmentlawoffoundationpit.【Keywords】foundationpitdeformation；filteringprocessing；chaostheory；deformationprediction；deformationlaw0引言伴随城市地下空间开发，基坑...
