摘要
对基于常规单一BP神经网络的电子鼻系统进行改进,提出一种基于Gabor原子神经网络的电子鼻系统,并以3种混合气体为实验对象,进行混合气体的定量分析研究。实验结果表明,应用Gabor原子神经网络的电子鼻系统的最大相对误差与单一BP神经网络相比得到减小,大大提高了定量分析精度。
A Gabor atom neural network used in electronic nose system is developed to improve the quantitative analysis of general BP neural network. Three mixed gases are adopted for the quantitative analysis. The result shows that recognition ability of the improved neural network is increased and the maximum relative error of quantitative analysis is decreased, the precision of quantitative analysis is increased.
出处
《电子技术应用》
北大核心
2011年第12期80-82,86,共4页
Application of Electronic Technique
关键词
电子鼻
神经网络
混合气体
模式识别
electronic nose
neural network
mixed gases
pattern recognition
