论文部分内容阅读
信号的特征提取和模式识别方法,在实现准确的电子鼻气体定性分析中尤为关键,本文提出了基于AR信号处理和KII模型的嗅觉识别算法.将传感器信号分为:上升期和稳定期两部分,对上升期信号提取斜率作为特征;对稳定期信号,进行AR建模来提取特征.在电子鼻的模式识别算法上,利用KII模型对气味信号进行分类.该方法充分利用了AR信号处理在信号表示方面的有效性及降维优势、KII模型在模式识别方面的优越性.仿真将该方法与BP网络、AR_BP算法及单KII网络进行了比较,结果表明,AR信号处理技术可以很好的提取特征,并与KII建立相关的数学模型,将AR信号处理技术应用到电子鼻系统中是可行的,且具有更高的识别率.
The method of signal feature extraction and pattern recognition is particularly crucial in the realization of accurate electronic nose gas qualitative analysis. In this paper, an olfactory recognition algorithm based on AR signal processing and KII model is proposed.The sensor signals are divided into two phases: rising phase and steady phase , The slope of signal ascending stage is taken as the feature, and the stationary phase signal is modeled by AR for feature extraction.In the electronic nose pattern recognition algorithm, the KII model is used to classify the odor signal.The method makes full use of AR signal processing in The effectiveness of signal representation and the advantage of dimensionality reduction, and the superiority of KII model in pattern recognition.Compared with BP network, AR_BP algorithm and single KII network, the simulation results show that AR signal processing technology can be very good It is feasible to apply AR signal processing technology to the electronic nose system to extract features and establish a mathematical model related to KII, and has a higher recognition rate.