Least-squares reverse time migration based on dynamically weighted dual conjugate parameters

Data-domain least-squares reverse time migration is an imaging method based on optimization theory that performs iterative optimization. In recent years, many researchers have continuously optimized the least-squares reverse time migration method, with the hybrid conjugate gradient method being the mainstream optimization algorithm for solving least-squares reverse time migration. However, the least-squares reverse time migration based on this method still suffers from issues such as slow convergence speed, leading to poor stability throughout the computational process. To address the above problems, this paper proposes a method that uses the Gauss-Newton direction as a guide and employs a weighted combination of dual conjugate parameters to improve the convergence stability of least-squares reverse time migration. The improved method dynamically weights and combines two gradient parameters, leveraging the fast convergence of the quasi-Newton algorithm and the stable convergence of the hybrid conjugate gradient method. Through testing on models and real data, compared with conventional methods, the proposed algorithm achieves better imaging results under the same number of iterations, while also demonstrating good stability and high computational efficiency.