摘要
本文根据1967～1982年间影响湛江站增水的热带气旋资料,选取了其中对湛江站增水影响比较显著的因子,运用人工神经网络中的BP算法,建立了湛江站风暴潮增水预报的人工神经网络模型,并使用贝叶斯优化算法提高了BP网络模型的预报能力,结果表明所建BP网络模型的训练拟合效果和预报效果都比较好.表明该网络模型可以用于湛江站风暴潮增水的预报.同时本文也为其它的港口、码头等近岸海洋工程风暴潮增水的预报提供了一个有效可行的方案.
According to the data of tropical cyclone which contributes to the surge tide level of Zhan Jiang station, we choose the key factors and use BP net of the artificial neural network to establish the storm surge forecast model of Zhan Jiang station. Then we use the method of Bayes regularization to improve the model’s forecast ability. The results show that both the effect of exemplar train fitting and the effect of forecast are well. That is to say the model we established has the ability to forecast the storm surge of Zhan Jiang station during the typhoon period. Meantime, we also provide a valid method to forecast the storm surge for other stations or habors.
出处
《海洋预报》
2005年第2期33-37,共5页
Marine Forecasts
基金
总装专题资助(40701010401)
