台风中心的自动定位对于灾害性天气预测具有重要意义。针对台风云系在云图中分布的不确定性和多变性,利用主特征向量提取方法实现了台风云系的客观描述;在此基础之上,根据台风密闭云区在云图中呈圆形的几何特性,构建了台风中心自动定位的非线性最优目标函数;因该目标函数的导数不易求取,利用生物遗传算法,求得了台风中心位置的最优数值解,有效地实现了台风中心的自动定位。多个台风中心定位仿真结果表明,该方法能够实现有眼和无眼台风的自动定位,并且理论可靠、实用性强、定位精度高。
It is important to locate the typhoon center automatically for forecasting destructive weather. Since the distribution of typhoon cloud bands from the satellite images is with uncertainty and irregularity, the dominant vector method is used to extract the typhoon features. A nonlinear optimal fitness function is set up according to the features of typhoon-cloud distribution. Based on the function, genetic algorithms are introduced to solve the fitness function, which locates the typhoon center automatically. Besides, the approach is applied to locate the center of other typhoons. Simulations show that the method proposed in this paper can locate the typhoon center with or without the eye. The method is practicable, feasible and highly accurate.