太赫兹衰减全反射技术对花生冻伤快速判别研究

Rapid identification of peanut frostbite based on terahertz attenuated total reflection technique

中文关键词:  花生冻伤  太赫兹衰减全反射技术  无损检测  随机森林

英文关键词:peanut frostbite  terahertz attenuated total reflection  nondestructive testing  random forest

中文摘要:

      目的 利用太赫兹衰减全反射(terahertz attenuated total reflection, THz-ATR)光谱法实现花生冻伤的快速鉴别。方法 实验选择种子公司购入的同品种冻伤和非冻伤花生各500粒, 采集1000粒花生样本的0~359.97 cm?1 THz光谱, 通过光学参数计算得到样本集的吸光度、折射率和吸收系数。采用3点移动窗口平滑预处理和随机森林算法建立基于不同光学常数的花生冻伤识别模型。结果 在决策树棵数为500, 特征变量数为38时, 基于太赫兹吸光度建立的花生冻伤判别模型性能最佳, 准确率、召回率、精确率达到97.0%、98.0%、96.1%。结论 本研究所建立的定性模型准确率高, THz-ATR技术有望为花生冻伤的快速无损鉴别提供一种新的高效的检测方法, 为太赫兹技术在食品检测领域的应用提供了现实依据。

英文摘要:

      Objective To realize the rapid identification of peanut frostbite by terahertz attenuated total reflection (THz-ATR) spectroscopy. Methods The 500 peanuts of the same variety of frost-damaged and non-frost-damaged peanuts purchased from seed companies were selected. The absorbance, refractive index and absorption coefficient of the sample set were obtained from the optical parameters calculated by selecting the 0?359.97 cm?1 THz spectra of 1000 peanuts collected for the experiment. The 3-point moving window smoothing and Random forest algorithm were used to establish a peanut frostbite recognition model based on different optical constants. Results The peanut frostbite discrimination model based on terahertz absorbance performed best with an accuracy, recall and precision of 97.0%, 98.0% and 96.1% at a decision tree number of 500 and a feature variable number of 38. Conclusion The qualitative model established in this study is highly accurate, and the THz-ATR technique is expected to provide a new and efficient detection method for the rapid nondestructive identification of peanut frostbite, providing a realistic basis for the application of terahertz technology in the field of food detection.