The Application of Probabilistic Neural Network Analysis Technique in the Prediction of Sedimentary Tuff Reservoir

Abstract

The tigh oil reservoir in Tiaohu group in Santanghu Basin has the characteristics of high-density porosity, ultra-low permeability and high oil saturation, which are mainly controlled by the sweet spot of reservoir.And, the prediction of lithology and effective porosity is very important in the study of sweet spot of tight oil. Probabilistic neural network (PNN) is a kind of neural networks based on probabilistic thinking and Bayesian classification rules. when geological lithology or seismic attributes changes or modifies, the network can automatically adapt to the new variables, and adjust the weights until convergence; we use PNN analysis technique to establish a nonlinear statistical relationship between seismic attributes and logging curve, and finally use this relationship to identify lithology and predict porosity. By applying of PNN analysis technique to predict lithology and porosity, we have confirmed that high porosity sedimentary tuff reservoir is mainly distributed in the north slope of the study area. In accordance with the results of reservoir prediction, all the proposed well sites have been drilled with high hydrocarbon flow from the Permian Tiaohu formation. The results show that the method is suitable for predicting sweet spot of the tight reservoir in the study area.