遗传算法在气动力参数辨识中的应用

Application of genetic algorithms for aerodynamic parameter estimation

将遗传算法推广用于气动力参数辨识,以取代通常采用的梯度类优化算法。通过采用遗传模拟退火算法对某型飞机的纵向气动力参数进行辨识计算及分析后,可以看到:(1)遗传算法是气动力参数辨识的一种新的有效方法,该算法不受参数初值选取的影响,具有较好的全局寻优特性;(2)遗传算法的计算效率受种群规模、遗传算法构造本身等因素的影响比较大,并且还有相当大的进一步完善与改进的空间。

Maximum Likelihood Estimation(MLE) method is the most prevailing method in aerodynamic parameter estimation. The most difficult problem of MLE is that the conventionallyused gradientbased optimization method in MLE relies too much on the selection of initial value of optimization. So, the Genetic Algorithms(GA) combined with the Simulated Annealing(SA) algorithm is applied in aerodynamic parameter estimation to replace those gradientbased optimization method. After implementing the GA to estimate the longitudinal aerodynamic parameters of an airplane, some preliminary conclusions can be drawn. First, GA is a new effective method to estimate aerodynamic parameters, and it is not restricted by the selection of initial values of parameters and able to find the globally optimal point effectively. Second, the efficiency of basic GA is greatly improved after combined with the SA technique thanks to the ability of SA to find the locally optimal point robustly. Moreover, the efficiency of GA is also influenced by other factors such as the number of population, etc., and there is a significant space for the further improvement and refinement of GA.