Performance Comparison of Optimization Algorithms in Matched Field Inversion

ZOU Shi-xin,YANG Kun-de,MA Yuan-liang
DOI: https://doi.org/10.3969/j.issn.1000-3630.2005.01.002
2005-01-01
Abstract:Optimization efficiencies and mechanisms of simulated annealing, genetic algorithm, differential evolution and downhill simplex differential evolution are compared and analyzed. Simulated annealing and genetic algorithm use a directed random process to search the parameter space for an optimal solution. They include the ability to avoid local minima, but as no gradient information is used, searches may be relatively inefficient. Differential evolution uses information from a distance and azimuth between individuals of a population to search the parameter space, the initial search is effective, but the search speed decreases quickly because differential information between the individuals of population vanishes. Local downhill simplex and global differential evolution methods are developed separately, and combined to produce a hybrid downhill simplex differential evolution algorithm. The hybrid algorithm is sensitive to gradients of the object function and search of the parameter space is effective. These algorithms are applied to the matched field inversion with synthetic data. Optimal values of the parameters, the final values of object function and inversion time is presented and compared.
What problem does this paper attempt to address?