Development of seismic inversion methods based on hybrid optimization of simulated annealing and quasi-Newton methods to estimate acoustic impedance and porosity from post-stack seismic data
摘要
This study proposes a hybrid optimization approach that integrates simulated annealing (SA) with the quasi-Newton method (QNM). SA is known for its ability to explore the solution space thoroughly and achieve a global optimum solution given sufficient computational resources and time. In contrast, QNM is a local optimization technique that can efficiently converge to a solution, but only if the initial model is sufficiently close to the global minimum or maximum. To address the limitations and leverage the strengths of both methods, this study introduces a unified framework that combines SA and QNM into a single, cohesive flowchart. The developed technique was tested first using synthetic data and a wedge model, and then it was used with real data from a Blackfoot field in Canada. The hybrid optimization method demonstrated excellent performance, delivering highly accurate inversion results with high resolution and a strong correlation between the original and inverted impedance and porosity. The additional statistical analysis such as mean, mode, standard deviation, correlation, and RMS error between real and inverted well data obtained after hybrid optimization produces quite excellent results. The correlation coefficients for the synthetic case, real impedance case, and real porosity case are 0.99, 0.84, and 0.59, respectively, and the RMS errors are 0.11, 0.26, and 0.36. From the inverted impedance and porosity sections, a low impedance (