改进的多层感知机对拟稳点的选取

Improvement Multi-layered Perceptron to Plans the Steady Spot the Selection

运用多层感知机对监测网中的相对稳定点(拟稳点)进行选取,克服了传统的剔除粗差的局限性,为监测网的拟稳平差提供精确可靠的拟稳点。对于多层感机不确定的隐层节点的数量做了适当的改进,应用到拟稳点的选取上,其数目可定。给出了一种适合于拟稳点选取的激励函数,而输出层的节点数,将其取决于检验显著水平。

This article utilizes in the multi-layered perceptron＇s（ to plan steady spot）to monitoring network relative point of fixity to carry on the selection, has overcome the traditional rejection thick difference limitation, plans the steady adjustment for the monitoring network to provide precisely reliable plans the steady spot. Has made the suitable improvement regarding multi-layered feeling machine indefinite implicit strata node＇s quantity, applies plans the steady spot in the selection, its number may decide. Gave one kind to suit in plans the steady spot selection the excitation function, but output the level the nodal point number,was decided it by the examination remarkable level