Print

Odor Sensing (Sniff out the future)

The electronic nose is a machine olfaction instrument that is to detect and discriminate complex odors using a sensor array. It has two important components in order to function: an electronic circuit to collect, transport odors to the sensor array and digitize sensor responses, as well as the signal processing and pattern analysis component.

In this lab, we focus on three different machine olfactory technologies: conducting polymer sensor (Cyranose 320), surface acoustic wave (SAW) sensor (fast GC analyzer), and colorimetric odor detector.

1. Conducting polymer sensor: Cyranose 320
The Cyranose electronic nose consists of 32 conducting polymer sensors and each sensor consists of three components: a substrate, a pair of gold-plated electrodes, and a conducting organic polymer layer. When the sensor is exposed to an analyte, the conducting organic polymer film swells, which causes the increase in resistance because the conductive pathways though the materials are disrupted. The pattern of the resistance change over the 32-sensor array (Figure 1) Figure 1. The Cyranose 320 and the 32-sensor array.provides the evidence for qualitative classification of different smell patterns.









We are investigating this electronic nose to detect Botrytis neck rot in Vidalia onions (Figure 2) and fungal diseases (Botrytis spp., Alternaria spp., and Colletotrichum spp.) in blueberries.

Figure 2. A typical Enose response to healthy onions and botrytis infected onions.

2. SAW sensor: fast GC analyzer
The fast GC analyzer consists of a surface acoustic wave (SAW) sensor (Figure 3), pneumatic controls and support electronics. It draws odors into the inlet via a pump.

Figure 3. The SAW sensor: faster GC analyzer. The volatile compounds pass through the valve and are adsorbed onto the trap. Then the absorbed compounds are vaporized by heating the trap and transported down to the capillary column where the compounds are separated based on their different solubility – this operation is similar to gas chromatography (GC) but it only needs roughly 10 seconds for sampling and thus it is called fast GC analyzer.
In this lab, we have successfully used this instrument for fungal diseases detection in red delicious apples (Figure 4).

Figure 4. A typical smellprint from the SAW sensor.













3. Advanced signal processing and Pattern recognition algorithms

Signal processing and pattern recognition algorithms are crucial for the success of the electronic nose sensing technology.

Figure 5. PCA analysis for healthy fruits and damaged fruits (Li et al., 2007).Since most electronic nose technology uses multiple sensors in an array for odor detection, its data are multivariate in nature and needs dimensionality reduction before further processing. Usually, principal component analysis (PCA) is used to extract features and reduce data dimension (Figure 5).



Figure 6. Electronic nose sensor reduction using the genetic algorithms (Li et al., 2008).Genetic algorithms and evolutionary strategy (ES) were successfully used for feature extraction in both Cyranose and fast GC analyzer (Figure 6). Pattern recognition algorithms used for electronic data analysis include discriminant analysis (DA), partial least squares (PLS), and cluster analysis (CA). The artificial neural networks (ANN) and supporting vector machine (SVM) usually achieve better performance than the statistical approaches.

References
Li, C. and P. Heinemann (2007). ANN integrated electronic nose system for apple quality evaluation. Transactions of the ASABE 50(6): 2285-2294.

Li, C., P. Heinemann and J. Irudayaraj (2007). Detection of apple defects using an electronic nose and zNose. Transactions of the ASABE 50(4): 1417-1425.

Li, C., P. Heinemann and P. Reed (2008). Using genetic algorithms (GAs) and CMA evolutionary strategy to optimize electronic nose sensor selection Transactions of the ASABE 51(1): 321-330.

Suslick, K. S., N. A. Rakow and A. Sen (2004). Colorimetric sensor arrays for molecular recognition. Tetrahedron 60: 11133–11138.


Project Leader: Changying "Charlie" Li
Contact Info:
cyli@uga.edu
Affiliation: University of Georgia
P.O. Box 748
Tifton, GA 31793
(229) 386-3170