基于遗传算法初始群体优化的台风中心自动定位

摘要

To solve the optimal fitness function of locating the typhoon center automatically using differential coefficient, it has the disadvantags of local optimal and optimal solution required difficulty. Genetic algorithm is applied to optimal locating the typhoon center automatically from IR satellite cloud images. It realizes the typhoon center automatic location. At the same time, considering the basic genetic operator has the problem of the initial population ran-domicity which can influence the optimal solution hunting. According to satellite meteorology theory, an improved genetic algorithm is proposed in terms of the creation of the initial population. Through the simulation of some typhoone centers automatic location, the proposed method shows reliability and rationality.%针对误差梯度求导方法求解台风中心自动定位最优目标函数时,具有解的局部优化问题以及最优解获取困难的不足,本文将遗传算法应用到卫星云图台风中心自动定位的优化求解中,实现了台风中心的准确自动定位.同时,考虑到标准遗传操作中初始群体的随机生成对最优解的搜索具有很强的敏感性和不确定性问题,本文根据台风云系的灰度分布特征,对初始群体的生成进行了改进与优化,求得了最优数值解.多个台风中心定位仿真试验结果验证了该方法的合理性和可靠性.