基于小波分解和最小二乘支持向量机的ENSO集成预测  

ENSO integration prediction based on wavelet decomposition and least squares support vector machine

作　　者：程亮[1] 刘家峻[1] 刘科峰[2] 余丹丹[1] 余运河[1] 

机构地区：[1]中国人民解放军61741部队 [2]解放军理工大学气象学院

出　　处：《海洋通报》2010年第4期367-371,共5页Marine Science Bulletin

摘　　要：用小波分解和最小二乘支持向量机相结合的方法,建立了ENSO的集成预报模型。该方法将复杂海温系统分解为相对简单的带通分量信号,然后建立分量信号的独立预报模型,最后对预报结果进行集成。试验结果表明,模型在保留预报对象主要特征的前提下,有效地降低了预报难度,集成预报准确率和预报时效均较传统方法有明显的改进和提高。Using the method of combining wavelet decomposition and least squares support vector machine, the integration prediction model of ENSO is established. With this method, the SST system is decomposed into a relatively simple band-pass component signals, then an independent forecasting model of component signals is set up, and the prediction results are integrated. The results show that while retaining the main features of the predicted objects, the model effectively reduces the difficulty of forecasting; integrated forecasting accuracy and forecasting timeliness are significantly improved and enhanced than that of the traditional methods.

关 键 词：小波分解 最小二乘支持向量机 赤道海温 

分 类 号：P732[天文地球—海洋科学]