基于高光谱技术和IRIV-FOA-ELM算法的花椒挥发油无损检测 下载: 958次
Nondestructive Testing of Volatile Oil of Zanthoxylum Bungeanum Based on Hyperspectral Technique and IRIV-FOA-ELM Algorithm
1 四川农业大学机电学院, 四川 雅安 625014
2 四川农业大学信息工程学院, 四川 雅安 625014
图 & 表
图 1. 果蝇优化算法优化极限学习机的流程图
Fig. 1. Fruit fly optimization algorithm optimized extreme learning machine flowchart
下载图片 查看原文
图 2. 标准正态变量变换预处理前后的光谱。(a)处理前(原始光谱);(b)处理后
Fig. 2. Spectra before and after SNVT processing. (a) Before processing (raw spectra); (b) after processing
下载图片 查看原文
图 3. 果蝇优化算法的适应度曲线
Fig. 3. Fitness curve of FOA
下载图片 查看原文
图 4. 优化前后测试集的对比。(a)优化前;(b)优化后
Fig. 4. Comparison of test sets before and after optimization. (a) Before optimization; (b) after optimaization
下载图片 查看原文
表 1训练集与预测集中花椒挥发油含量统计表
Table1. Statistics of pepper oil content in training set and prediction set
Sample | Number ofsamples | Pepper oil content /(mL·g-1) |
---|
Minimum | Maximum | Mean | Standard deviation |
---|
Total | 150 | 0.03 | 0.08 | 0.056353 | 0.011984 | Training set | 120 | 0.03 | 0.08 | 0.056108 | 0.012253 | Prediction set | 30 | 0.031 | 0.075 | 0.057333 | 0.010978 |
|
查看原文
表 2基于迭代保留信息变量算法建立模型的建模结果
Table2. Modeling results of different models based on IRIV
Model | Calibration set | Prediction set |
---|
| RMSEC | | RMSEP |
---|
Primitive variables-PLS | 0.5714 | 0.6922 | 0.4933 | 0.7565 | All variables-PLS | 0.6157 | 0.6702 | 0.5614 | 0.7477 | IRIV-PLS | 0.6732 | 0.5944 | 0.6362 | 0.6524 | All variables-ELM(17) | 0.7409 | 0.7219 | 0.6952 | 0.8057 | IRIV-ELM(17) | 0.8547 | 0.4311 | 0.6557 | 0.6059 | All variables-ELM(18) | 0.8312 | 0.5872 | 0.7522 | 0.6700 | IRIV-ELM(18) | 0.8522 | 0.3475 | 0.8365 | 0.5737 | All variables-ELM(19) | 0.8425 | 0.4840 | 0.6862 | 0.5014 | IRIV-ELM(19) | 0.8446 | 0.4121 | 0.6952 | 0.6688 | All variables-ELM(20) | 0.8392 | 0.6126 | 0.7494 | 0.6965 | IRIV-ELM(20) | 0.8546 | 0.4673 | 0.7609 | 0.5994 |
|
查看原文
表 3FOA-ELM模型的结果
Table3. FOA-ELM model results
Model | Calibration set | Prediction set |
---|
| RMSEC | | RMSEP |
---|
All variables-FOA-ELM(18) | 0.7573 | 0.5170 | 0.7004 | 0.5939 | IRIV-FOA-ELM(18) | 0.8792 | 0.3323 | 0.8659 | 0.3621 |
|
查看原文
纪然仕, 陈晓燕, 刘素珍, 饶利波, 汪震. 基于高光谱技术和IRIV-FOA-ELM算法的花椒挥发油无损检测[J]. 激光与光电子学进展, 2020, 57(20): 203002. Ranshi Ji, Xiaoyan Chen, Suzhen Liu, Libo Rao, Zhen Wang. Nondestructive Testing of Volatile Oil of Zanthoxylum Bungeanum Based on Hyperspectral Technique and IRIV-FOA-ELM Algorithm[J]. Laser & Optoelectronics Progress, 2020, 57(20): 203002.