Neural network-based ground stress field inversion and fault influence study in south Sichuan shale gas reservoirs

Several faults are developed in the Luzhou area of South Sichuan. The influence of faults on the ground stress field is complicated. A finite element numerical model was established based on geological and logging data. The stress field in the study area was inverted using an optimised neural network algorithm. The distribution of the stress field and the influence of faults on the ground stress field were studied. The results show that: The accuracy of the neural network algorithm reaches more than 90%; The maximum horizontal principal stresses range from 112 to 148 MPa. The minimum horizontal principal stresses range from 90.3 to 122 MPa. The stress regime is maximum horizontal principal stress > vertical stress > minimum horizontal principal stress; The influence of the tectonic faults on the stress field is mainly divided into geometric factors and material factors. When the fault material parameter (modulus of elasticity) is close to the strata parameter, it mainly affects the minimum horizontal principal stress. When the modulus of elasticity of the fault decreases, the effect of the fault on the maximum horizontal principal stress increases; The smaller the modulus of elasticity of the fault, the greater its influence on the effect of well stress. The results of the study provide a reference for the inversion of the stress field and the prediction of fault cracking in shale reservoirs in the Luzhou area.