Realizing genetic algorithm of optimal log interpretation

Genetic algorithm is a novel optimizing method with the character of global optimization and it is suitable for solving the problems of non-linear optimization. In recent years, it has been successfully applied to many fields, such as optimal solution of complex function, structure optimal design, adaptive control, system control and pattern recognition, etc. The genetic algorithm was introduced into optimal log interpretation by the authors, in other words, the optimal log interpretation was realized by use of the genetic algorithm. The questions about multi-parameter and nonlinear optimum log data processing were studied by fully applying the character of global optimization of the genetic algorithm. On the basis of fully discussing conventional genetic algorithm theory, it is pointed out in the paper that the conventional genetic algorithm was improved from two aspects: 1 introducing non-time-homogeneous genetic algorithm in light of the severe non-convergence of the conventional time-homogeneous genetic algorithm; and 2 in combination with integral simulation annealing algorithm, introducing dynamic record window technique in order to solve the problem of pre-mature convergence. Through forward model simulation operation and practical log data interpretation, it is shown that the genetic algorithm improved is much easier to be converged, by which the multi-parameter and nonlinear optimum log data interpretation may be well studied.