Research Of The Wing Aerodynamics Optimization Technology Based On A Improved Genetic Algorithm

Aircraft aerodynamics configuration optimization is aimed to obtain betteraerodynamics characteristic by adjusting the geometrical configuration of aircraft,it is one kind of an important nonlinear system containing large quantity ofvariables. Comparing with the traditional optimization algorithms based on thegrads which only get a local optimal solution, genetic algorithm as a bionicoptimization algorithm, has a stronger robust and nature parallel. What is more,genetic algorithm also has better global solution capability when dealing withcomplicated nonlinear system. However, when we use the standard geneticalgorithm on this optimization problems directly, the convergence ration becomesrather slowly, even unreachable, because the aircraft aerodynamics configurationoptimization contains numerous variables, and the N-S equations are not easy tobe solved and require so much time.In order to enhance the efficiency genetic algorithms, there were someimprovement on the standard genetic algorithm and a new genetic algorithmwhich is specially suited for aerodynamics configuration optimization was set upand has been used on airfoil and wing aerodynamics configuration optimization.The following work is performed:1. Some improvement was made on the standard genetic algorithm in thispaper. Based on the real-coded genetic algorithm, there were mainly someimprovement on the crossover operator and mutation operator which affect thegenetic algorithms capability obviously, a new cross operator which can guide thepopulation evolving to a better direction and also a new mutation operator thatadjusts the mutation probability automatically based on the distance ofindividuals and also the evolvement process were designed in this paper.Otherwise, RSSR operator, double population strategy, elitist strategy and nichestrategy were also brought in by referring some successful improvements on thegenetic algorithm which have been proved successful.2. It is hard to set the parameters of GA and the GA parameters influence thecapacity of GA obviously. In this paper, a detailed study on the GA parameters,such as crossover operator probability, mutation operator probability, nicheradius and so on was done, and a guide line for setting GA parameters was givenfor most normal problems. The numerical test proved that the guide line is usefulfor most problems.3. A fast elitist non-dominated sorting genetic algorithm (NSGA-2) formed for dealing multi-objective optimization problems which was based on theimproved genetic algorithm seen above and the fast elitist non-dominated sortingmethod.4. Introducing the Bezier-Bernstein function to parametric the airfoilconfiguration and an in-depth research on Bezier-Bernstein function was done inorder to investigate the influence on numbers and range of design variables.5. Based on the natural parallel capacity of genetic algorithm, a parallelgenetic algorithm and a fast elitist non-dominated sorting genetic algorithm(NSGA-2) software for aerodynamic configuration optimization wereaccomplished by combining the grid morph part and flow solver on the Linuxclusters through the PBS system using the FORTRAN95computationallanguage.The numerical tests show that the convergence of improved geneticalgorithm enhanced obviously, and the improved genetic algorithm can deal withcomplicated aerodynamics configuration optimization. The genetic optimizationmodel module works well with gird morph module and flow solver module, andthe optimization platform based on improved genetic algorithm has a primaryability in the actual project application.