摘要
用BP算法，采取两个惯性项的惯性调整权重系数方法和新的学习步长调整法，对乙烷氟氯衍生物的常压沸点作预测，根据预测出的沸点数据指出了CFC11，CFC12和CFC113的一些可能的代用品。由于采用新的训练策略，网络的收敛速度得到提高．
The back-propagation algorithm in artificial neural networks is improved in this paper. The convergence speed can be increased with two inertial terms for adjustment of weights and new control scheme of learning step. The improved algorithm is applied in prediction of boiling points of the fluorine and chlorine-substituted ethane derivatives. Some possible alternatives for CFC11, CFC12 and CFC113 can be found with the prrdicted boiling point values.
出处
《计算机与应用化学》
CAS
CSCD
1997年第1期65-67,共3页
Computers and Applied Chemistry
关键词
人工神经网络
BP算法
沸点
乙烷氟氧衍生物
Artificial neural network,Back-propagation algorithm, Fluorine-chlorine-substituted ethane, Boiling point prediction, Alternatives for CFC
