| 基于人工神经网络的肉类光谱识别 |
| |
| 引用本文： | 马钉凌,张铁强.基于人工神经网络的肉类光谱识别[J].光谱实验室,2007,24(5):914-917. |
| |
| 作者姓名： | 马钉凌 张铁强 |
| |
| 作者单位： | 吉林大学南岭校区逸夫大楼物理教育中心,长春市人民大街,130025 |
| |
| 摘 要： | 利用自组的光纤探头式光谱仪对载体肉类表面进行光谱测量,采用反向传播人工神经网络(BP-ANN)对肉类可见光谱进行识别,分析了隐层神经元、期望误差、判别输出范围等网络参数的调整对识别功能的影响.基于大量实测数据样本,优化网络结构参数选择,建立了良好的人工神经网络,对肉类光谱识别率达到97.5%.
|
| 关 键 词： | 光谱分析 肉类识别 BP神经网络 参数优化 |
| 文章编号： | 1004-8138（2007）05-0914-04 |
| 修稿时间： | 2007-06-27 |
The Study on the Effect of the ANN's Parameters on the Identifying Meat Spectrum |
| |
| MA Ding-Ling,ZHANG Tie-Qiang.The Study on the Effect of the ANN''''s Parameters on the Identifying Meat Spectrum[J].Chinese Journal of Spectroscopy Laboratory,2007,24(5):914-917. |
| |
| Authors: | MA Ding-Ling ZHANG Tie-Qiang |
| |
| Institution: | College of Physics, Nanling Campus of ,Jilin University, Changchun 130025, P. R. China |
| |
| Abstract: | The method using the artificial neural net was used to identify the reflected spectra of different meats,which were measured with fiber sensor spectrometer.Particularly,the effects of different numbers of hidden layers,different MSE goals,different ranges of output on the ANN were studied in detail.Based on volumes of samples,the ANN's Parameters were optimized,and the identifying ratio for tested samples is up to 97.5%. |
| |
| Keywords: | Spectrum Analysis Meat-Species Identification BP-ANN Optimize Parameters |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
