摘要
提出了采用人工神经网络方法来重点预测多元混合气体的爆炸上限,并将其与偏最小二乘线性回归、Le Chatelier方法相比较。仿真结果表明,BPNN的预测结果远好于PLSR以及Le Chatelier计算出的结果,由此表明BPNN对混合可燃气体的爆炸上限具有更好的预测和泛化能力。
A A method of back-propagation (BP) artificial neural network for predicting the upper flammability limits of fuel mixtures is proposed. The simulation result shows that the average relative error of BPNN is less than that of Partial Least-Squares Regression Method(PLSR) and the result calculated by Le Chatelier. It suggests that this method has better prediction and generalization ability for the upper flammability limits of the fuel mixtures.
出处
《现代化工》
CAS
CSCD
北大核心
2016年第2期159-160,162,共3页
Modern Chemical Industry
基金
国家自然科学基金项目(61203072
61203133)
江苏省六大人才高峰项目(DZXX-042)
关键词
安全工程
爆炸上限
混合气体
BP神经网络
偏最小二乘回归
safety engineering
upper flammability limits
fuel mixtures
back-propagation neural network (BPNN)
partial least-squares regression