基于小波分析与BP神经网络智能化识别基桩缺陷.pdfVIP

基于小波分析与BP神经网络智能化识别基桩缺陷 吴治涛1 宋玉强1 张磊1 刘 伟1 1．四川省煤田地质局物探工程院四川成都610072 摘要：本文通过分析传统数字滤波方法处理动测信号的缺点，提出了基于小波分析与神经网络智能化识别基 桩缺陷的思路。利用bior5．5小波基把非平稳动测信号分解为6个不同层次的频带序列，在有效尺度信号中提取 基桩缺陷的特征向量，然后结合BP神经网络建立了神经网络智能化识别基桩缺陷的模型，开发了基桩缺陷智 能化识别软件FPIC。结果表明小波分析更加精细的刻画了动测信号的内部信息，BP神经网络对基桩缺陷的智 能化识别精度较高，此方法及软件对于提高基桩动测信号的分析水平具有实用价值。 关键词：基桩缺陷；小波分析；BP神经网络；智能化识别 ClassificationoftheFoundationPileDefectBasedon Intelligent 何匀velet andBPNeuralNetwork analysis WUZhi-taolSONG ZHANGLeii UUWeii Yu—qiangi Coalfield 0072 1．Geophysicalinstitute，SichuanGeologyBureau，Chengdu，Sichuan，61 engineering Abstract：Thisarticle the oftraditional methodtodealwiththe test forwardthe analysis dynamicsignal，put shortage di醇alfiltering ideaswhichiscombined‘thewavelet andneuralnetwork identifieationdefect．the test callbe analysis intelligent pile dynamicsignal．If basis different valuecanbeextractedfor bior5．5intosix leveland band decomposedby’wavelet frequencysequence，thepowerspectrum some band identificationmodelCanbeestablishedwavelettransformfeatureextractionand sequence．The by specified intelligent using BPartificialneural writethesoRwareFPICto classificationconclusionshowthatthe test network,we intelligent pile．The dynamicsignal insideinformationCanbemore wavelet defectsoffoundationcanbeidentifiedtheBPartificialneural flayedby analysis，the pile by networkwith and methodandsoftwarehave