基于PCA和BP混合神经网络的无损电子鼻设计

The design of undamaged E-nose base on hybrid neural network of PCA and BP

本文基于近红外光谱吸收技术检测原理,设计并开发出了一套能够快速、高效、准确的对目标气体进行检测的电子鼻系统。整套电子鼻系统主要包括近红外激光发射单元,气室单元,探测器单元和基于。LabVIEW的数据处理单元在数据处理单元中,)PCA(Principal Component Analysis,主成分分析)算法和BP(Back Propagation,反向传播神经网络通过。Lab VIEW所提供的MATLAB Script节点,集成到上位机软件中,并用来对采集到的数据进行分析实际的测试中,该电子鼻被用来对气体样品中的HCl、CH4和NH3成分进行检测,测试结果表明该电子鼻系统可以对样品中的目标气体成分进行准确识别。总体而言,整套电子鼻系统运行稳定可靠,检测速度快,识别准确度高,能够很好的满足设计预期。

An e-nose based on near infrared spectral absorption technology which can be used to detect target ga_ses accurately and efficiently was designed and implemented in this paper . The e-nose was composed of near-infrared laser emission unit, gas chamber unit, detector unit and data processing unit based on LabVIEW. PCA （ Principal Component Analysis） recognition algorithm and BP （ Back Propagation） neural network which were used to analysis the collected dater were integrated into the upper computer software by the MATLAB Script node supplied by Lab_VIEW. In the testes, this e-nose was used to detect HCl、CH4 and NH3 in the gas samples. The result indicated that the designed e-nose can identify the target gases without error based on prior study. Generally speaking, the e-nose designed in this paper was stable and reliable and could detect the target gases effectively with high accuracy. And the design met the design objectives very well.