Sinopec Develops High Accuracy Intelligent Seismic Inversion with Deep Learning

MATLAB has played an important role in Sinopec’s research, specifically when it comes to using Signal Processing Toolbox™, Optimization Toolbox™, and Statistics and Machine Learning Toolbox™ for geophysical modeling.

Key Outcomes

  • Analyzed frequency-phase characteristics of seismic data
  • Developed an innovative frequency-phase intelligent inversion method
  • Achieved remarkable improvement in practical application
Video length is 19:40

Sinopec is one of the largest oil and gas companies in China. Its business involves the exploration, extraction, storage, and transportation of oil and gas.

One of the main challenges in petroleum exploration is meeting high-precision requirements for seismic inversion. Traditional methods of seismic inversion that rely on geological models are sensitive to noise and hence often produce low-resolution results. At Sinopec, engineers are using MATLAB® to apply data-driven techniques and solve this problem.

Sinopec has introduced a new seismic inversion method called frequency-phase intelligent inversion, which combines seismic frequency-phase features and deep learning. Using MATLAB, engineers extract seismic wavelets, conduct high-resolution time-frequency analysis, construct massive data label pairs, and optimize and train deep networks.

Using frequency-phase intelligent inversion, high-resolution impedance inversion results can be obtained for both simulated and real seismic data.