基于广义回归神经网络的β型磷建筑石膏强度预测

Prediction of β-phosphogypsum Strength Based on General Regression Neural Network

本文利用工业废弃物磷石膏制备β型磷建筑石膏,并确定了影响β型磷建筑石膏强度的因素及特点,在此基础上,建立了β型磷建筑石膏强度预测的广义回归神经网络(General Regression Neural Network,GRNN)模型,利用实验室中制备β型磷建筑石膏的15组统计数据作为学习样本,通过网络拟合训练和预测分析,得到了较高精度的预测结果,证明了GRNN的非线性映射能力、容错性和自学习性用于β型磷建筑石膏强度预测是非常有效的,避免了大量盲目的配比试验及资源浪费,提高了实验水平和效率。

Phosphogypsum from industrial waste is used to preparing β-phosphogypsum. The General Regression Neural Network （GRNN） model for the prediction of β-phosphogypsum strength is established by determining the influent factors and characteristics of β-phosphogypsum. The network is trained with 15 groups laboratorial data as learning sample, which gets a good fitting effect. With the high precision of predicted results, the ability of GRNN which includes nonlinear mapping function, fault tolerance and self-study, is demonstrated efficaciously in predicting β-phosphogypsum＇ s strength. By using GRNN model , large amounts of repeated proportioning test and resource waste can be avoided, improve economic benefits.