Application of an Artificial Intelligence Method for Velocity Calibration and Events Location in Microseismic Monitoring

Good quality hydraulic fracture maps are heavily dependent upon the best possible velocity structure. Particle Swarm Optimization inversion scheme, an artificial intelligence technique for velocity calibration and events location could serve as a viable option, able to produce high quality data. Using perforation data to recalibrate the 1D isotropic velocity model derived from dipole sonic logs (or even without them), we are able to get the initial velocity model used for consequential events location. Velocity parameters can be inverted, as well as the thickness of the layer, through an iterative procedure. Performing inversion without integrating available data is unlikely to produce reliable results; especially if there are only one perforation shot and a single poor-layer-covered array along with low signal/noise ratio signal. The inversion method was validated via simulations and compared to the Fast Simulated Annealing approach and the Conjugate Gradient method. Further velocity model refinement can be accomplished while calculating events location during the iterative procedure minimizing the residuals from both sides. This artificial intelligence technique also displays promising application to the joint inversion of large-scale seismic activities data.