基于电子鼻和人工神经网络的沉香与沉香曲鉴别

Identification of Aquilariae Lignum Resinatum and Chenxiangqu Based on Electronic Nose and Artificial Neural Network

目的:利用沉香和沉香曲气味的不同实现对两者的鉴别。方法:通过单因素考察和正交试验建立适合的电子鼻检测方法,对沉香和沉香曲的气味进行检测。将样品分为训练组和验证组,利用训练组的电子鼻检测原始数据和归一化数据训练建立径向基函数(RBF)神经网络、广义回归神经网络(GRNN)和概率神经网络(PNN),以验证组正确率为指标对人工神经网络进行参数优化。结果:经参数优化后的3种不同人工神经网络对验证组沉香和沉香曲的识别正确率均可达100%。GRNN在不同数据预处理方式和不同参数条件下对验证组的正确率均达100%。结论:电子鼻和人工神经网络技术可利用沉香和沉香曲气味的不同对两者进行有效区分。

Objective:To study the identification of Aquilariae Lignum Resinatum and Chenxiangqu by their different odor.Methods:The detection method of odors of Aquilariae Lignum Resinatum and Chenxiangqu using electronic nose was established by single factor method and orthogonal experiments.And samples were divided into training set and test set.Then the raw data and normalized data of the training set were used to train radial basis function(RBF),generalized regression neural network(GRNN)and probabilistic neural network(PNN).The parameters of artificial neural networks(ANNs)were optimized by using the classification accuracy of test set as the study index.Results:With optimized parameters,the classification accuracy of test set of all three ANNs were 100%.And the classification accuracy of GRNN was 100%with different data pretreatment and different parameters.Conclusion:Electronic nose and artificial neural network technology can be used to distinguish Aquilariae Lignum Resinatum and Chenxiangqu.