Intelligent optimization based prestack inversion method for high resolution estimation of elastic parameters

Seismic inversion plays a vital role in constructing high-precision subsurface models for reservoir characterization, fault detection, tunnel engineering, and dynamic formation analysis. To improve inversion accuracy and computational efficiency, this study proposes a novel indirect gradient-based pre-stack amplitude variation with angle (AVA) inversion method, termed mayfly algorithm (MA)-based nonlinear AVA inversion (MANAI). The method leverages the MA within a Bayesian framework and utilizes an accurate Jacobian matrix derived from the exact Zoeppritz equations. The core innovation lies in the indirect global optimization strategy, which extracts gradient information from a prior model before initiating the meta-heuristic optimization. This approach eliminates the need to compute reflection coefficients at every iteration, significantly reducing computational cost while maintaining inversion accuracy. To validate the method's effectiveness, we compare the MANAI approach with both the whale optimization algorithm (WOA)-based inversion and a conventional local nonlinear method. A parallel implementation is also developed to further accelerate the inversion process. Benchmark tests on synthetic data demonstrate the superior stability and accuracy of the proposed method under varying conditions. Application to real seismic data further confirms that MANAI delivers improved inversion results and reduced runtime compared to traditional strategies. These findings highlight the method's potential for efficient and reliable inversion in complex exploration scenarios.