<p>Aiming at the higher requirements for inversion methods in seismic processing and interpretation, as well as the lack of precision control means for the search process in the Genetic-Simulated Annealing algorithm, this paper proposes a new strategy. Here we construct an inversion strategy based on the idea of dynamic statistical description of Markov chains. By embedding the Genetic Algorithm (GA) into the internal iteration loop of Simulated Annealing (SA), a nested iteration mode is formed, providing two adjustment factors at different scales to enhance the robustness of the optimization iteration process. Under the Bayesian framework, this strategy can give the quantitative uncertainty of the inversion results. The two - level precision control acts on the crossover and mutation strategy of the inner loop (GA) and the cooling strategy of the outer loop (SA) respectively, jointly and finely controlling the optimization process of the GA - SA combined algorithm. Through the test analysis of mathematical models and the application of actual data, compared with the conventional GA-SA search algorithm, the new method has significant advantages in inversion accuracy. The inversion results are reliable and stable. Meanwhile, its quantitative uncertainty description helps processing interpreters understand the confidence and multiplicity of solutions of the inversion results, providing a more effective solution for seismic inversion.</p>