The Research And Application Of Co-evolutionary Algorithm In Meteorological Field

There are a large number of numerical optimization problems in meteorological field, such as:model, equation parameter optimization problems and numerical forecasting data assimilation problems. Evolutionary algorithm using mathematical models to modeled these problems in order to solve the problem itself, which has important theoretical value and practical significance. Coevolutionary algorithm is a new class of evolutionary algorithms which based on the theory of co-evolution:It considered the coordination between population and the environment, between the population and the population in the evolutionary process, and thus to ensure the diversity of population and improve the efficiency of evolution. In this paper, For the characteristics of co-evolution algorithm, designed a double elite populations co-evolutionary algorithm and apply the algorithm to the numerical optimization problems in meteorological field, Firstly, the algorithm for air quality assessment model parameters are optimized, According to the measured concentration data samples for air quality assessment done research; Secondly, according to the Lorenz equations combined with three-dimensional variational, applied the coevolutionary algorithm to weather forecast model data assimilation done related research.This paper can be devided into the following three aspects in detail:1. This paper conducted an exploratory development relative to the existing problems of the Fast non-dominated sorting genetic algorithm (NSGA-II) and propose a double elite populations coevolutionary algorithm.There are some limitations in NSGA-II:uneven distribution of population convergence, weak global search ability,algorithm run slower,In this algorithm:The two elite populations using different evolutionary strategies;A variety of crossover and mutation operators. All of these can balance the capability of exploitation and exploration.2. The proposed algorithm is applied to the evaluation of the model parameter optimization in the atmosphere and improved air quality assessment results. According to the air quality damage rate formula established objective function, and using double elite populations co-evolutionary algorithm to optimization objective function. This paper uses the measured data as an evaluation sample, the use of air quality assessment model algorithm to optimize this paper based on the air quality after the evaluation, application of this algorithm to verify the effect of the meteorological model optimization.3. The proposed algorithm is applied to three-dimensional variational data assimilation method, improved data assimilation effect. Accoring to the three-dimensional variational data assimilation method to establish the objective function.The proposed algorithm is applied to three-dimensional variational method, The experimental results show that three-dimensional variational data assimilation methods based on this algorithm has good performance and verify the algorithm in numerical forecasting is effective and practical.In summary, this paper aiming at the shortage of NSGA-II algorithm, design a co-evolutionary algorithm based on double elite populations. The algorithm provides an effective solution for the optimization problem in this field. Work in this paper not only enriches the study of co-evolution algorithm, but also to further expand the application fields of collaborative algorithm.