关键词: 四川盆地, 沙溪庙组, 气藏, 流体, 密度, 神经网络, 反演

Abstract: The fluid-density changes from rich gas reservoir to water layer obviously, so it is a nice parameter of reservoir gas forecasting. This paper presents the method for calculating the fluid-density values of each reservoir using density logging and porosity data, according to the density-porosity least-square inversion method, calculated the fluid-density of reservoir in well. The inversed three dimensional fluiddensity results can be obtained based on the probabilistic neural network inversion method, and using the well fluid-density data and three dimensional seismic data. Application of this method to Shaximiao gas reservoir in a gas field of western Sichuan area has achieved very good results, providing important data support for reservoir prediction, gas-water identification and gas reservoir description in this area.

Key words: Sichuan, fluid, density, neural network, inversion