用人工嗅觉系统评定卷烟的内在品质

迄今为止，卷烟的内在品质是靠评吸专家的感官的评定的，由于受到其经验，情绪以环境条件等多种因素的影响，评定结果的准确性难以保证。为此，作者运用模仿人的嗅觉系统功能的人工嗅觉系统来评定卷烟的内在品质。深入研究了人工嗅觉系统的气敏传感器阵列和数据处理分析方法，提出了阵列的组成方法以及神经网络集和测试集的构造方法。

ARTIFICIAL OLFACTION SYSTEMS FOR EVALUATING THE INTERNAL QUALITY OF CIGARETTES

Up to now,the internal quality of cigarettes has been evaluated by sense organ of men who are evaluating experts in quality of tobaccos. Since the evaluating results are bound to be affected by experts' mood and experience, environment conditions and so on, the accuracy of evaluating results is hard to be guaranteed. In this paper, an artificial olfaction system, which is used to imitate the function of human olfaction system,is applied to evaluate the internal quality of cigarettes. Two important parts of the system, that is gas sensor array and data processing apparatus, are researched in depth. Meanwhile, the arranging method of the array and the producing method of the training set and the testing set of RBF neural network are presented. The feasibility of evaluating the internal quality of cigarettes with artificial olfaction system was proved by an experiment and its data processing results.