基于概率神经网络的产品质量预测方法研究

A Study of the Methods for Prediction of Product Quality Based on Probabilistic Neural Networks

对通过正交试验法采集样本数据、采用概率神经网络(PNN)对产品质量进行预测的方法进行研究,并将预测结果与反向传播神经网络(BP)预测结果进行比较。结果表明,在利用正交试验法采集的训练样本数据分别训练PNN和BP网络后发现,概率神经网络的预测准确度远远大于反向传播神经网络,显示了概率神经网络在模式识别领域的优势。

In this paper, both back-propagation neural networks（BPNN） and probabilistic neural networks（PNN） are applied to prediction of product quality, and the simulated results show that when the samples training BPNN and PNN are filtered by orthogonal analysis, the success rate of PNN is much higher than that of BPNN. The superiority of PNN is presented in the pattern recognotion.