基于神经网络的油墨转移率预测

利用神经网络根据油墨转移率影响因素的数据来预测其具体数值. 通过对比径向基神经网络、Elman神经网络和BP神经网络最后选择径向基神经网络作为预测网络的模型. 根据各影响因素间的相互关联和各自对油墨转移率影响的大小关系确定影响油墨转移率的主要因素. 以主要影响因素为试验条件,运用正交试验法和均匀试验法合理设计分组试验,将试验结果作为神经网络的样本数据. 利用样本数据对网络进行训练,最终使网络能够预测出不同影响因素下油墨转移率的数值. 将径向基神经网络预测结果和Elman神经网络、BP神经网络的预测值对比分析,以此证明径向基神经网络的预测值具有较高的精确度.

Prediction for Ink Transferring Ratio Based on Neural Network

In this paper, the specific values of the ink transferring ratio was predicted according to the data of its influential factors by using neural network. RBF neural network was selected as the model of prediction network by comparing the radial basis function ( RBF) neural network, Elman neural network and back propagation ( BP) neural network. The main influence factors of ink transferring ratio were determined according to the correlation among the various factors and their impact on the ink transferring ratio. Considering the main influencing factors, a group of experiments were taken by using orthogonal experimental design and uniform design experimentation, and the result was used as the sample data of the neural network. The RBF neural network was trained by the sample data, which enable the network to predict the ink transferring ratio under different factors. Results show that the predicted value of RBF neural network is more accurate than Elman neural network and BP neural network through contrastive analysis.