Artificial Olfactory System

Prepared By P.ANITHA

10/24/09

Artificial Nose

1

Contents What is E-Nose Components Working Principle Sensors Available Data Analysis Difference between human nose and E-Nose Application Conclusion Reference

10/24/09

Artificial Nose

2

Electronic Nose Developed in early 1980’s. Electronic Nose is a smart instrument that is designed to

detect and discriminate among complex odors using an array of sensors. It is a device that combines gas sensors, electronics and

pattern recognition software to sense odors present in the environment, their type and in some cases concentration.

10/24/09

Artificial Nose

3

Components of ENose Chemical Sensing System An array of several different sensing elements Odors samples are sensed and patterns named and

kept in the database Database trains the pattern recognition system

Goals Configures the recognition system Automatic identification of chemicals

Example Spectrometer

10/24/09

Artificial Nose

4

Components of ENose Pattern recognition system  Analyze complex data  Recognize patterns  Unknown chemicals can be rapidly identified in the

field

Example Artificial neural networks

10/24/09

Artificial Nose

5

Working Principle

10/24/09

Artificial Nose

6

Working Principle Chemical sensor Odor samples are exposed to gas sensors The sensor array “sniffs” the vapors from a sample and provides a set of measurements Pattern Recognizer Recognize new patterns The pattern recognizer compares the pattern of the measurements to stored patterns for known materials Perform qualitative or quantitative analysis

10/24/09

Artificial Nose

7

Working Principle Output  When stabilization is reached the e-nose emits a

signal and the substance is identified,results are shown in a pc

A cleaning process is to be done for each time A small pump drives “ clean ” synthetic air

through the chamber until the sensors and the chamber itself are clean and ready to start. After that a micro-electro-valve switches to inject air coming through the inlet hose, which will be near the odor source. 10/24/09

Artificial Nose

8

Algorithms Backpropagation-trained Feed-forward networks Fuzzy ARTmaps Kohonenos self-organizing maps (SOMs) Learning vector quantizers (LVQs) Hamming networks Boltzmann machines Hopfield networks

10/24/09

Artificial Nose

9

Different types of sensors Metal Oxide Sensors Whose resistances vary with the type and quantity of gas absorption Surface Acoustic Wave sensors whose acoustic properties are modified by gas

absorption

Conducting Polymers odor sensitive substances whose resistance change upon gas absorption 10/24/09

Artificial Nose

10

10/24/09

Artificial Nose

11

10/24/09

Artificial Nose

12

Data Analysis Principal Component Analysis Multilayer Perception SIMCA Partial Least square Principal component regression Cluster analysis

10/24/09

Artificial Nose

13

PCA Reduces the high multidimensionality of a

complex data set into two or three variables –the principal components Contain the majority of the information necessary to discriminate the samples More of a Statistical Method Store all Data as an m x n matrix (A)

10/24/09

Artificial Nose

14

10/24/09

Artificial Nose

15

10/24/09

Artificial Nose

16

Artificial Neural Network The most powerful type of data processing

technique being employed in Electronic Nose instruments ANNs are self-learning; the more data presented, the more discriminating the instrument becomes ANNs allow the Electronic Nose to function in the way a brain functions when it interprets responses from olfactory sensors in the human nose 10/24/09

Artificial Nose

17

How Neural Nets Will Benefit the Electronic Nose Sensor responses drift over time Change with concentration Sample Depletion

Relate the “Pattern” of the data Find the data that is the closest Determine the odor

10/24/09

Artificial Nose

18

Why is an Electronic Nose Better? Trained human ‘sniffers’ are expensive Individuals vary Hazardous Chemicals Can be done in real time for long periods

10/24/09

Artificial Nose

19

Problems where E-Nose help Best suited for matching complex samples with

subjective endpoints such as odor or flavor The E-Nose can match a set of sensor responses to a calibration set produced by the human taste panel or olfactory panel routinely used in food science Example when has milk turned sour when is a batch of coffee beans optimally roasted

10/24/09

Artificial Nose

20

Analogy Between the Electronic Nose and the Biological Nose Inhaling ⇒ Pump Mucus ⇒ Filter Olfactory Epithelium ⇒ Sensors Binding With Proteins ⇒ Interaction Enzymatic Reactions ⇒ Reaction Cell Membrane Depolarized ⇒ Signal Nerve Impulses ⇒ Circuitry & Neural Network

10/24/09

Artificial Nose

21

Difference Human Nose

Electronic Nose

 array of hundreds of

sensors  organic sensors  broad band capability  trainable to new odors  data acquisition in the brain  analysis by true Neural Network  processing; pattern recognition 10/24/09

Artificial Nose

 array of a few tens of sensors  thin film polymer based

sensors  broad band capability  polymers selected to respond to  particular compounds  trainable to new analytes  data acquisition by computer  data analysis by computational  methods and pattern recognition

22

Limits to Human Nose Fatigue Odor adaptation Insensitivity to some species Toxicity of some contaminants

10/24/09

Artificial Nose

23

Advantages of E-nose Portable High Reliability Identify simple molecules

10/24/09

Artificial Nose

24

Limits of E-Nose Difficulty in maintaining an exhaustive database of

different fingerprints of chemicals. It may be very difficult to analyze a complex mixture of different chemicals. The precision of the device while analyzing similar smell is controversial. Insensitivity to some species High cost Time Delay

10/24/09

Artificial Nose

25

Applications Food Industry Medicine Environment Cosmetic, perfumery Chemistry Etc.,

10/24/09

Artificial Nose

26

Food Industry Application This is currently the biggest market for

electronic noses: Inspection of food to test for ripening/rotting. Testing of packaging materials for odour containment. Verifying if orange juice is natural. Grading whiskey and controlling fermentation.

10/24/09

Artificial Nose

27

Medicine  Sensitive `noses' may help clinicians by

examining Breath odours:

Abnormal breath can be indicative of

gastrointestinal problems, sinus problems, infections, diabetes and liver problems.

Body fluids:  The smell of urine and blood can help in the diagnosis of

liver and bladder problems

Wounds: early diagnosis of wound infection considerably improves healing rate 10/24/09

Artificial Nose

28

Environmental Monitoring Monitoring of factory emissions, air quality

and household odours. Detection of oil leaks. Analysis of toxic wastes and fuel mixtures.

10/24/09

Artificial Nose

29

Cosmetic, perfumery Quantification monitoring with E-Nose (SAS-

FMS) - Evaluation of fragrances in shampoo Chemistry Paper odors Oxidized oils odors Solvents from paint or plastics

10/24/09

Artificial Nose

30

Conclusions The enose offers some clear advantages over

other techniques currently in use to determine odur levels It is particularly useful in routine operations due to its ease of use and rapid response rate Humans are not well suited for repetitive or boring tasks that are better left to machines The E-nose has the interesting ability to address analytical problems that have refractory to traditional analytical approaches 10/24/09

Artificial Nose

31

References

 http://www.asctbhopal.com/Technovista07/FEVERTONICS/5/elec%20smell%  A paper presentation on ANN for enoseby Y.Sarika and N.Sambhavi  The Electronic Nose: aquick and dirty introduction  Electronic Nose by Sree Sailaja  EN by Paisan Doungjak  The Electronic Nose & its potential by Miss Asha Joseph,Dr. Peter

Lykos  Applications and Advances in Electronic-Nose Technologies by Alphus D. Wilson 1, Manuela Baietto  http://www.pdfcoke.com/doc/6775979/Enose  http://cns-web.bu.edu/pub/laliden/WWW/Papers/nose.html

10/24/09

Artificial Nose

32