油菜菌核病预报方法的研究

STUDY ON FORECASTING OF SCLEROTIA SCLEROTIUM

在滑动相关普查的基础上,分别建立以气象、海温和环流资料为预报因子的油菜菌核病病情指数预报子模式。对子模式采用平均、加权、回归和人工神经网络(ANN)方法进行综合集成。结果表明,集成模式提高了历史样本的拟合精度和独立样本试报的准确性,特别是人工神经网络集成模式的效果更令人满意。

In terms of sliding correlation analysis, three sub models of sclerotia sclerotium were set up, based on meteorology, air circulation and sea surface temperature predictors respectively, With the aid of these sub models, four synthesis ensemble prediction models were established by the way of average, weighted mean, regressing and artificial neural networek(ANN). Comparative results show that the ensemble forecasting models, especially the ANN one, are superior in historical fittings and testing prediction compared to the sub models.