Application of conventional logging interpretation fracture method based on neural network in offshore Oilfield L

Abstract

The well data is required as a constraint for fracture prediction in the area. At present, imaging logging is considered to be the most accurate and effective means to interpret fractures, which can show the geological characteristics of the two-dimensional space of the borehole wall intuitively, vividly and clearly, but the experimental data is scarce due to the high cost of measurement. Conventional logging data are the main logging data in domestic oil and gas fields, so it is very important to use conventional logging data to effectively interpret fractures in the absence of drilling coring and imaging logging data. Based on the neural network algorithm, taking offshore Oilfield L in China as the application target, this paper optimizes the combination of conventional logging curves by establishing the nonlinear mapping relationship between the fracture strength curve and the conventional logging curve of sample wells with imaging logging data. By establishing the neural network model with nonlinear high learning ability for the optimized conventional logging curve and fracture strength curve, the fracture identification is carried out for the target interval of non sample wells. The results show that this method can combine geology, geophysics, mathematics, computer and so on, and can effectively identify fractures in the study area. It has good adaptability to offshore fractured reservoir.