Multi-attribute probabilistic neural network inversion applicated in identifying igneous in RWP area based on cross-validation

There developed huge thick Permian Igneous in Tabei area of Xinjiang.The Permian igneous rocks with sharp variation of velocity,affects the process of oil and gas exploration seriously,and makes trap-confirming more difficult.For solving this problem,this paper use Probabilistic Neural Network inversion method to establish igneous velocity field.Compared with CSSI,PNN inversion is a typical nonlinear inversion with its high resolution.At first,a group of attributes was selected by using Stepwise regression and cross-validation for analyzing and error minimum,to make inversion results have better correlation with log properties.The inversion velocity field was testified to conform the distribution of igneous and velocity changes.