鞭角华扁叶蜂滞育幼虫空间数量分布的研究

Study on the Spatial Quantitative Distribution of Chinolyda flagellicornis

运用人工神经网络方法和空间趋势面分析方法对鞭角华扁叶蜂滞育幼虫的数量与其所处的空间位置之间的关系进行研究 ,创建鞭角华扁叶蜂滞育幼虫空间数量分布的 BP网络模型和三次趋势面模型。结果表明 :建立的 BP网络模型模拟不同空间位置的鞭角华扁叶蜂滞育幼虫数量具有较高的拟合精度 ,当隐层神经元数为 12个时 ,其平均模拟精度为 96.90 % ,预留的 2 0组样本的平均预测精度为 92 .35 % ;而三次趋势面模型的平均模拟精度为 93.0 1% ,预留的 2 0组样本的平均预测精度为 93.2 4 %。

The artificial nerve network method and spatial tendency plane analysis method were used to study the relationship between the amount and spatial position of Chinolyda flagellicornis diapause larva.The BP (Back Propagation)network model and cubic tendency plane model of the spatial quantitative distribution of C.flagellicornis diapause larva were established.The results showed that the BP network model has high precision of fit in simulating the amount of diapause larvae with different spatial position.When the amount of crypto neurone is 12,the mean simulation precision reaches 96 90%.The mean forecast precision of 20 groups of sample prepared in advance reachs 92 35%.The mean simulation precision of cubic tendency plane model is 93 01%,while the mean forecast precision of 20 groups of sample prepared in advance is 93 24%.