The application of improved genetic algorithm in fitting the spatial variogram

The implementation of automatically fitting the theoretical model of spatial variogram has always been a problem in the realm of geostatistics. Considering that the genetic algorithm has sufficient robustness and it has the virtue of searching the global optimum while solving the nonlinear problems, so this paper takes the method of genetic algorithm to automatically fit the theoretical model of spatial variogram. However, the traditional genetic algorithm has the deficiency of slow convergence rate and gets premature convergence easily, according to which this paper proposes a kind of new improved genetic algorithm: comprehensively considering the optimal preservation, self-adaptive regulated probability of intercross and of mutation, introducing the simulated annealing algorithm to revise the fitness function and the survival mechanism. The validation of typical cases and application in fitting the spatial variogram show that this improved algorithm converges quickly and is globally convergent. It is worthy to say that the improved genetic algorithm is a simple and effective method of fitting spatial variogram and other nonlinear problems.