基于广义回归神经网络的阳极效应自动预测

Anodic Effect Automatic Prediction Based on Generalized Regression Neural Network

针对当前阳极效应预测方法存在精度低、过拟合等缺陷,根据阳极效应的非线性、时变性等变化特点,设计了一种基于广义回归神经网络的阳极效应自动预测模型。首先采集阳极效应预测的样本,并对样本数据预处理,建立阳极效应预测的学习样本,然后将阳极效应的学习样本输入到广义回归神经网络进行学习,构建阳极效应自动预测模型,最后进行了具体的阳极效应预测仿真实验,并与其他模型进行了阳极效应预测的对比测试。结果表明,广义回归神经网络可以有效的拟合阳极效应变化特点,提高了阳极效应预测精度,而且预测误差明显要小于当前其他阳极效应预测模型,具有较高的实际应用价值。

Aiming at the defects of low accuracy,over fitting and so on,a new prediction model of anode effect based on generalized regression neural network is designed based on the characteristics of anodic effect,such as nonlinearity and time variation.Firstly,anode effect prediction samples are collected,and sample data pretreatment to establish anode effect prediction learning samples,and then learning sample of anode effect are input to generalized regression neural network to learn and construct the automatic prediction model of anode effect,simulation experiments of anode effect prediction is carried out and compared with other test models.The results show that the proposed model can effectively fit the variation characteristics of anode effect to improve the prediction accuracy of anode effect,and predict error is significantly less than other current anode effect prediction models,so it has high practical application value.