节点文献
一种改良BP网络方法研究
A STUDY ON MODIFIED BACKPROPAGATION NEURAL NETWORK
【摘要】 误差反向传播神经(BP)网络存在学习速度较慢、易陷入局部极小等缺点。通过自适应步长、反传改进参数γ、数据预处理和权值初始化等改良BP网络,使标准BP算法收敛速度大大加快,收敛平稳性获得提高。方法用于钙、镁、铁的三组分混合物同时定量分析,取得一定效果。
【Abstract】 A new modified backpropagation neural network method is proposed based on a novel varying steplength, modified network parameters and data pretreatment. The learning and the convergent stability of this new network have increased. The method is used for simultaneous analysis of Ca, Mg and Fe elements in synthetic samples with satisfactory results
【关键词】 改良BP网络;
钙;
镁;
铁;
同时分光光度法;
【Key words】 modified Backpropagation neural network; Ca; Mg; Fe; simultaneous spectrophotometry;
【Key words】 modified Backpropagation neural network; Ca; Mg; Fe; simultaneous spectrophotometry;
- 【文献出处】 桂林工学院学报 ,JOURNAL OF GUILIN INSTITUTE OF TECHNOLOGY , 编辑部邮箱 ,1998年04期
- 【分类号】TP18
- 【被引频次】5
- 【下载频次】54