摘要
利用浙江海岛站1天4次的平均风速作为人工神经网络的输入层，相对应的日极大风为输出层，建立日极大风速人工神经网络预报模型，试报表明夏季大风天气系统和秋冬季大风天气系统的预报拟合率为89％～90％，春季大风天气系统的预报拟合率为87％～88％，夏季大风天气系统和秋冬季系统6～9级极大风预报准确率较高，基本在70％左右，对预报员有较好的参考价值。与以前大风预报方法相比，它具有简单、有效的优点，可以不受地域限制在各地推广使用。
By using mean wind velocity data of four times per day as input layer, and corresponding maximum wind velocity as output layer, the daily maximum wind velocity forecasting modes of neural network were built. The trials of forecasting show that the correlation coefficients between forecast results and real data with summer gale system and autumn-winter gale system reach 89% to 90%, and 87% to 88% with spring gale system. The forecasting accurate rates of six to nine grdae maximum wind velocity with summer gale system and autumn-winter gale system are higher, which reach 70% and have good referencing value to forecasters. Compared with other forecasting methods of gale, it features simplicity and effectiveness and could be applied in various regions without restriction.
出处
《海洋预报》
2006年第B09期64-67,共4页
Marine Forecasts
关键词
人工神经网络
极大风预报
准确率
artificial neural network： forecasting of maximum wind velocity accurate rate
