<p>This study introduces a novel music-inspired hybrid optimization approach for simultaneous pre-stack seismic inversion, aimed at enhancing the accuracy and efficiency of subsurface property estimation. The proposed method, termed hybrid harmony search optimization (HHSO), combines the global search ability of harmony search optimization (HSO) with the local refinement efficiency of the quasi-Newton method (QNM). While HSO has shown effectiveness in global optimization, its convergence rate can be slow. The integration with QNM addresses this limitation by accelerating convergence and refining the inversion results more effectively. The HHSO algorithm is evaluated using both synthetic and real seismic datasets. For synthetic noise levels up to 30%, HHSO delivers significantly improved estimations of acoustic impedance (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">${Z}_{p}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Z</mi> <mi>p</mi> </msub> </math></EquationSource> </InlineEquation>), shear impedance (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">${Z}_{\text{s}}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Z</mi> <mtext>s</mtext> </msub> </math></EquationSource> </InlineEquation>), and density (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\rho$</EquationSource> <EquationSource Format="MATHML"><math> <mi>ρ</mi> </math></EquationSource> </InlineEquation>) compared to HSO and QNM methods. Application to real data also shows consistent improvements, with HHSO achieving higher correlations to well-log data (<InlineEquation ID="IEq4"> <EquationSource Format="TEX">${Z}_{p}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Z</mi> <mi>p</mi> </msub> </math></EquationSource> </InlineEquation> = 0.91, <InlineEquation ID="IEq5"> <EquationSource Format="TEX">${Z}_{\text{s}}$</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>Z</mi> <mtext>s</mtext> </msub> </math></EquationSource> </InlineEquation> = 0.89, and <InlineEquation ID="IEq6"> <EquationSource Format="TEX">$\rho$</EquationSource> <EquationSource Format="MATHML"><math> <mi>ρ</mi> </math></EquationSource> </InlineEquation> = 0.92) and reducing inversion errors from 1.4 to 0.25. Moreover, HHSO preserves the frequency content of the seismic signal more effectively, which is crucial for maintaining the integrity of seismic data. When applied to full seismic sections, HHSO produces smoother and more continuous images of the subsurface, aiding in the identification of geological features such as layer continuity and sand facies zones. These findings highlight the potential of HHSO as a robust and efficient tool for seismic inversion and reservoir characterization, offering an optimal balance between global exploration and local precision.</p>

A music-driven hybrid optimization approach for simultaneous pre-stack seismic inversion: a case study

期刊论文
发表日期：

2026年4月28日

查看全文

Ravi Kant,
Md Sultanul Arefin,
S. P. Maurya

摘要

This study introduces a novel music-inspired hybrid optimization approach for simultaneous pre-stack seismic inversion, aimed at enhancing the accuracy and efficiency of subsurface property estimation. The proposed method, termed hybrid harmony search optimization (HHSO), combines the global search ability of harmony search optimization (HSO) with the local refinement efficiency of the quasi-Newton method (QNM). While HSO has shown effectiveness in global optimization, its convergence rate can be slow. The integration with QNM addresses this limitation by accelerating convergence and refining the inversion results more effectively. The HHSO algorithm is evaluated using both synthetic and real seismic datasets. For synthetic noise levels up to 30%, HHSO delivers significantly improved estimations of acoustic impedance ( ${Z}_{p}$ $Z_{p}$ ), shear impedance ( ${Z}_{\text{s}}$ $Z_{s}$ ), and density ( $\rho$ $ρ$ ) compared to HSO and QNM methods. Application to real data also shows consistent improvements, with HHSO achieving higher correlations to well-log data ( ${Z}_{p}$ $Z_{p}$ = 0.91, ${Z}_{\text{s}}$ $Z_{s}$ = 0.89, and $\rho$ $ρ$ = 0.92) and reducing inversion errors from 1.4 to 0.25. Moreover, HHSO preserves the frequency content of the seismic signal more effectively, which is crucial for maintaining the integrity of seismic data. When applied to full seismic sections, HHSO produces smoother and more continuous images of the subsurface, aiding in the identification of geological features such as layer continuity and sand facies zones. These findings highlight the potential of HHSO as a robust and efficient tool for seismic inversion and reservoir characterization, offering an optimal balance between global exploration and local precision.