红外云图的台风内核风速建模的RBFNN 和PDE 方法

目前反演台风内核风场时多采用线性回归方法进行建模,针对基于线性回归法的台风内核风速拟合效果较差的缺点,提出一种基于径向基函数神经网络(RBFNN)和偏微分方程(PDE)结合的红外卫星云图有眼台风内核风速和云图灰度建模方法。首先采用基于测地活动轮廓模型的PDE提取有眼台风的眼壁,获得台风眼壁空间位置和亮度数据;然后结合台风年鉴给出的台风近中心最大风速数据基于RBFNN进行有眼台风内核风速和云图灰度建模。实验结果表明:该算法改善了台风内核风速拟合效果,算法性能优于传统的线性回归法。

Typhoon inner core wind speed modeling method by RBFNN and PDE based on infrared cloud image

At present, linear regression model is often used to estimate typhoon inner core wind field. But the fitting effect of typhoon inner core wind speed based on linear regression was bad. Based on infrared satellite cloud image, radial basis function neural network (RBFNN) and partial differential equation(PDE) were used to build a model between typhoon inner core speed and cloud image's gray value. Firstly, typhoon's eye wall was extracted by using PDE which based on geodesic active contour model from the infrared satellite cloud image and the eye wall's space position and brightness are obtained. Then the maximum wind speed near typhoon center which was recorded by typhoon yearbook was used to build a model between typhoon inner core's speed and cloud image's gray value by RBFNN. The experimental results show that the proposed algorithm improves the fitting effect of typhoon inner core's wind speed, and the overall performance of the proposed algorithm is better than tradition method of linear regression.