摘要
以实船板水火加工成型的测试数据为样本，应用改进的ＢＰ网络，分别建立了帆型曲板的最大收缩长度、收缩面积的７参数和５参数的４个神经网络模型． 介绍了网络的结构设计及其训练过程，进行了网络的误差分析；算例结果证明该模型具有实用意义．
Based on the improved BP neural network, four models of 7 parameters and 5 parameters with regard to the maximum contraction length and the contraction area are established by using experiment data of practical ship plate bending by line heating. The network structure and the training process are explained, and the error resulted from the models is discussed.
出处
《大连理工大学学报》
CAS
CSCD
北大核心
1999年第6期797-801,共5页
Journal of Dalian University of Technology
基金
辽宁省博士科研起动基金
辽宁省自然科学基金
