抄録
Full Waveform Inversion (FWI), as a key technology for high-resolution seismic imaging and parameter estimation, has undergone rapid development in oil and gas exploration in recent years. Driven by advances in multi-scale acquisition systems, high-performance numerical simulation, and the expansion of intelligent algorithms, FWI has achieved continuous breakthroughs in multi-component, multi-mode, and multi-scale joint inversion, enabling an effective transition from theoretical refinement to engineering practice. Meanwhile, emerging approaches such as deep-learning-assisted inversion, the integration of physical priors, data assimilation, and uncertainty quantification have significantly enhanced the stability and efficiency of inversion workflows, fostering the formation of a “data-driven–physics-driven” hybrid paradigm. However, FWI still faces key bottlenecks—including strong nonlinearity with multiple minima, sensitivity to the initial model, noise amplification, and high computational costs—which constrain its broader applicability. Looking ahead, the intelligent evolution of FWI is advancing toward coupled frameworks that incorporate digital-twin simulation, generative models, and quantum computing, aiming to build high-precision inversion systems with adaptive optimization and multi-source data integration. This review systematically synthesizes the theoretical evolution, technical challenges, and intelligent development pathways of FWI, providing a comprehensive reference and guidance for high-fidelity imaging and intelligent inversion in oil and gas exploration.
