BP神经网络非线性组合预测模型在海洋冰情预测中的应用

针对海洋冰情灾害的非线性复杂问题，目前已提出了多种模型对其进行预测。在此基础上，根据神经网络的非线性和良好的函数逼近特性，提出用基于BP神经网络的非线性组合预测（NN-NLCF）模型来预测海洋冰情灾害。结果表明，NN—NLCF模型与海洋冰情的非线性特性相契合，它综合利用了参与组合的多种预测模型的有效信息，因而能更客观地反映海洋冰情的发展趋势，预测结果更为稳健、精度更高，在其它自然灾害时序预测中具有一定的推广应用价值。

BP Neural Network Based Nonlinear Combination Forecasting Model for Forecasting Marine Ice Condition

Several models have been presented to forecast marine ice condition disaster according to its nonliear complex problems.In this paper,nonlinear combination forecasting(NN-NLCF) model based on neural network isapplied to forecasting marine ice condition disaster,which is nonlinear and has excellent character of function approximation.The result shows that owing to being consistent with nonlinear character of marine ice condition and taking full advantage of effective information of combined models,NN-NLCF model can reflect the objective trend of marine ice condition,its forecasting is robust and accurate relatively,and it can be widely applied to forecasting different natural disaster time series.