摘要
针对复杂神经网络模型的结构特性和ＢＰ网络梯度下降法的缺点，将遗传算法与神经网络相结合，提出了基于改进连续化遗传算法的神经网络学习方法，并将其应用于大型乙烯裂解过程产品分布预测。
According to the structure characteristics of neural networks, this paper proposes a neural networks learning algorithm based on continuous genetic algorithm. By making use of continuous genetic operators, the algorithm can directly probe the solution in state space avoiding the process of coding and decoding, and can improve the searching performance. A numerical example of predicting the products distribution of gas oil pyrolysis shows that the improved algorithm is effective.
出处
《化学反应工程与工艺》
EI
CAS
CSCD
北大核心
1998年第3期329-333,共5页
Chemical Reaction Engineering and Technology
基金
国家自然科学基金重点资助项目
关键词
神经网络
遗传算法
乙烯裂解
生产
模拟
Neural networks, Continuous genetic algorithm, Gas oil pyrolysis
