东海区域灾害性海浪长期预测方法研究  

Study on the Long-term Predicting Way of Disastrous Sea Wave of East China Sea

作　　者：王华[1] 姚圣康[1] 龚茂珣[1] 陈美榕[1] 

机构地区：[1]国家海洋局东海预报中心,上海200081

出　　处：《海洋通报》2007年第5期35-42,共8页Marine Science Bulletin

基　　金：国家海洋局青年海洋科学基金资助项目(编号:2005217)

摘　　要：海浪预报具有复杂性、非线性和高维性等特点,而BP人工神经网络是处理非线性关系,进行科学预测行之有效的方法。本文中采用BP人工神经网络技术,以1985—2005年影响东海区的热带气旋、温带气旋和冷空气等相关要素为样本,建立东海区域灾害性海浪BP神经网络长期预测数值模型。通过模拟预测效果检验:该模型预测准确率超过85%,平均误差约为8%,满足实际预测要求。The characteristics of sea wave forecast are complexity, nonlinearity, high-dimension and soon. Being effective in handling nonlinear problems, the BP artificial neural network can carry out scientific forecasting. Using the data of tropical cyclones, temperate zone cyclones and cold air from 1985 to 2005 which affecting the east China Sea area, we adopt the BP artificial neural network technology, and set up a numerical forecasting model of disastrous sea wave of the east China Sea. By examining the results, the model＇s accurate rate exceeds 85 %, and the average error is about 8 %, which meet the needs of the real forecasting operation.

关 键 词：BP人工神经网络 灾害性海浪 隐含层 神经元 

分 类 号：P732.7[天文地球—海洋科学]