冰情预报的投影寻踪回归模型

将投影寻踪回归模型应用于黑龙江上游江段开河日期的预报,并与GA-BP模型预报的结果进行了对比分析.结果表明,投影寻踪回归模型预报的精度及稳定性较高,其性能优于常用的GA-BP模型.预报采用理论分析与多元逐步回归分析相结合的方法筛选预报因子,既可确保不遗漏基本影响因子,又能剔除对目标值影响不显著的因子,用于确定冰情预报的预报因子较为适宜.

Projection pursuit regression model for ice situation forecast

This paper applies projection pursuit regression(PPR) model to break-up date of the upstream of Hei-long River and gives a comparative analysis with GA-BP model. The result indicates that the PPR model's forecast accuracy and stability are comparatively high and whose working performance is superior to the common GA-BP model. Combining theoretical analysis with multivariate stepwise regression analysis method to select forecasting factors in the forecasting can not only ensure the basic factors completely but also cut out the unessential factors to target value, which is comparatively appropriate to determination of the forecast factors in ice situation forecast.