利用误差逆传播神经网络法识别几种毒品的太赫兹光谱

在采用太赫兹时域光谱(THz-TDS)技术对9种常见毒品进行实验研究并得到它们在0.2～2.6THz频率范围的特征吸收谱的基础上,用误差逆传播(BP)神经网络法对9种常见毒品的太赫兹吸收光谱进行了训练及识别。首先,用9种毒品的太赫兹吸收谱训练已经建立的误差逆传播神经网络;然后,选用与训练光谱不同时间测得的9种毒品的太赫兹吸收光谱作为检测光谱,经过二阶导数预处理之后分别输入到训练好的误差逆传播神经网络中进行识别,识别率达到89%。该误差逆传播神经网络模型采用MATLAB语言编制程序。识别结果充分表明,用误差逆传播神经网络可以实现对不同种类毒品的识别和鉴定,为太赫兹光谱技术用于毒品的检测和识别提供了一种有效的方法。

Identification of Terahertz Absorption Spectra of Illicit Drugs Using Back Propagation Neural Networks

On the base of absorption spectra in the range from 0.2 to 2.6 THz of nine illicit drugs using terahertz time-domain spectroscopy(THz-TDS) technique,the THz absorption spectra of the nine different illicit drugs were identified successfully by back propagation(BP) neural networks.Firstly,absorption spectra of the nine illicit drugs,which were pretreated by second derivation,were used to train the BP neural network.Secondly,absorption spectra of the nine illicit drugs which were measured in different time,pretreated by second derivative too,were identified by BP neural network and the identification rate of 89% was achieved.The model of BP neural network was processed in Matlab.The results indicated that it is feasible to apply BP neural network model on the identification of illicit drugs,and providing an effective method in the secure inspection and identification for illicit drugs using THz-TDS technique.