测井岩性的非线性动态辨识新方法及应用

由于岩性测井曲线分布具有模糊性,在对岩性进行划分时会出现较大的困难.为了准确分析测井响应曲线,将逃逸微粒群算法与Elman反馈神经网络进行有机结合,形成了EPSO-NN混合算法,并构建了基于"EPSO-NN"的非线性动态识别系统,用于测井岩性的自适应识别.工程实例结果表明,该系统在岩性识别上是可行的、有效的,同样也完全可以用于岩相、沉积微相识别、矿床预测及矿物岩石分类地质方面的研究.

New method and application of dynamic identification in well logging lithology

Fo r the fuzzy dist ribut ion of w ell log ging curves, it is very diff icult to par tit ion litho logy . In orderto analyze the well logg ing response curves in the process of lithology ident if icat io n, escape part iclesw arm opt imizat io n ( EPSO) w as proposed to be combined w ith Elman neural netw ork ( ENN) , and EPSONNhy brid alg orithm w as pro duced. T he nonlinear dy namical ident if icat ion system w as const ructed basedon EPSO-NN. T he system w as employed to adapt ively ident ify the w ell logg ing litholog y. The result s ofthe engineering cases indicate that this sy stem is feasible and ef fect iv e in the litholo gy ident if icat io n, so itcan be completely used in the g eo logical researches such as ident ificat ion of litholog y, lithofacies and sedimentarymicro facies, ore depo sit predict ion and mineral rock classification.