About the meeting: The „OLFACTORY BIORESPONSE III meeting” is the third conference in a series of meetings which started in 1995 at the Department of Pharmacology at the University of Erlangen, Germany. The two previous meetings of this series of conferences have been received extremely well by all participants, largely because a major focus is on the interpersonal exchange between researchers. The scientific focus of the meeting is on studies using electrophysiological and imaging techniques. Among other topics the 2003 meeting is going to highlight retronasal
39
Embed
Zostały wykonane w ramach magisterskiej pracy dyplomowej mgr inż. Beaty Krajewskiej
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
About the meeting: The „OLFACTORY BIORESPONSE III meeting” is the third conference in a series of meetings which started in 1995 at the Department of Pharmacology at the University of Erlangen, Germany.
The two previous meetings of this series of conferences have been received extremely well by all participants, largely because a major focus is on the interpersonal exchange between researchers.
The scientific focus of the meeting is on studies using electrophysiological and imaging techniques. Among other topics the 2003 meeting is going to highlight retronasal olfactory perception, olfaction in neurodegeneration, and qualitative olfactory dysfunction.
Zostały wykonane w ramach
magisterskiej pracy dyplomowej mgr inż. Beaty Krajewskiej
(nagroda II stopnia w Konkursie Ministra Środowiska "Nauka na rzecz ochrony środowiska i przyrody"
na najlepsze prace magisterskie przygotowane w polskich szkołach wyższych w 2003 roku)
Badania były jednym z etapów projektu badawczego:
„Intensywność zapachu. Prawa psychofizyczne i sztuczne sieci neuronowe”
(2001-2003; kierownik pracy: dr hab. inż. J. Kośmider).
Poniżej zamieszczono prezentację przygotowaną przez mgr inż. Beatę Krajewską
na OLFACTORY BIORESPONSE III,OLFACTORY BIORESPONSE III,
a po konferencji przedstawianą na Seminarium Doktoranckim WTiICh PS w języku polskim
(patrz – notatki prelegenta)
Joanna Kośmider, Beata Krajewska
Odour Monitoring Adopting GC-NN method
Technical University of Szczecin, Department of Chemical Engineering and Environmental Protection Processes, Laboratory for Odour Quality of the Air
Dresden Olfactory Bioresponse 2003
1. Introduction
2. Research methodology:a) sampling,
b) chromatographic analysis,c) sensory analysis,
d) artificial neural network application
3. Results of the researches
Plan of the presentation
4. Conclusions
INTRODUCTION
A property of a chemical compound or of mixtures of compounds depenent on the concentration to activate the sense of smell and then be able to start an odour sensation
Introduction
An individual sensation dependent on sensibility of human olfactory analyser and motivational factors
Odour - definition
Introduction
Each compound:• volatile in the conditions of the surroundings,• dissolvable in water,• dissolvable in fat,
• of eligible amount of molecules in the air (eligible concentration S),
• polar, while contacting protein receptors stimulating olfactory cells,
induces odour sensation of intensity I.
Introduction
I = k W–F · log (S/SPW)
Trials of combining strength of sensation (odour intensity,I) with strength of stimulus (odorant concentration, S),
psychophisical functions:1. Weber – Fechner law
I – strength of sensation (intensity), [ - ],kW–F – coefficient of proportionality (Weber – Fechner coefficient), [ – ],S – strength of stimulus (odourant concentration in air inducing odour sensation of intensity I), [mg/m3],SPW – odour sensation threshold, [mg/m3].
I = ks · S n
2. Stevens law
I – strength of sensation (intensity), [ - ], S – strength of stimulus (odourant concentration in air inducing odour sensation of
intensity I), [mg/m3],kS, n – empirical constants, [ – ].
Introduction
Odour– definition reffering to both pleasant and unpleasant olfactory sensations
Legal restrictions on odour emissions
Trials of regulating problems with odour quality of the air have been undertaken in different countries for more than 30 years:1. Japan (since 1972),2. Canada (Quebec, since 1980),
4. Germany,5. Poland.
3. Holland (since 1984),
Introduction
The most unambiguous and complex description of the problem was prepared by German legislation:Restrictions on odour emissions reffer to all industrial works irrespective of whether they are subject to the procedure of sanctioning their activity or not (different ways of executing the restricions in various regions).
The most advanced trials of regulating problems with odour difficulties - North Westphalia:
Guideline ‘Odours immission’ – frequency of occurance exceedings the threshold concentration of olfactory detectability of air pollutants (so called ‘odour hours’).
Introduction
Operational use of area Immission standardIW
habitable and miscellanous 0,1trade and industrial 0,15
Limiting frequencies of odour hours occurance (Germany)
Share of negative estimations in the total number of estimations
Prescriptions of German Agricultural Department
Operational use of soil
Limitary quantities
Odourants concentration
[ou/m3]
Frequency of evceedings
[% h in a year]
habitable areas 1 3miscellanous
dedication 1 5
rural areas 1 83 3
industrial areas1 10
3 5
Project of polish olfactory standard of air quality determines the highest admissible concentration elaborated by our researching group:
Type of area Class of hedonic quality of air
Time of exceedings TON30
= 0,1 ou/m3
[%h in a year]habitable and recreational
H0 5H1 3
industrial and rural
H0 10H1 8
• H0 – neutral or pleasant odour,• H1 – unpleasant odour
Introduction
Introduction
a. to determine olfactory difficulty of polluted air, especially in industrial areas where odourants emissions are much higher than in any others (with sensory analysis of samples of polluted air),
b. to verify the determined quantities with the standarised threshold values,
2. Air consists of mixtures of odourants to which quoted psychophisical laws are not applicable on the contrary to isolated compounds...
A fact that:
...provoked the idea of applying GC-NN system to evaluate odour intensity of mixtures of
compounds.
1. It is essential:
Aims of the work
1. Verifying potentiality of artificial neural networks to predict odour intensity of mixtures of compounds,
2. Determining existence of correlation between a feature of odour quality – odour intensity, I and 14 values describing the sample responsible for the odour (14 distinctive points of a chromatographic curve measured [mm] from an invariable basis, h1 - h 14)
3. Determining magnitude of training sets for ANN to achieve the best results (the smallest error measured with SD. RATIO, RMS Error and irrelative error)
Introduction
Introduction
SIMILARITY
ANNBIOLOGICAL NEURAL NETWORK
SIMILARITY
RESEARCH METHODOLOGY
Research methodology
Sensory analysis
Artificial neural network
Odour intensity,
I1
Chromatographic data, h1 - h14
SET OF DATA
Odour intensity obtained with
analitical methods, I2
Chromatographical analysis
• taking samples of pure air
Research methodology
a. SAMPLING
Stroehlein Gas Cylinder
Accumulatore
Heat-resistant foil sleeve
Polietylen hose
Materials
Research methodology
SAMPLING
• taking samples of pure air
Stroehlein Gas Cylinder
Accumulatore
Heat-resistant foil sleeve
Polietylen hose
• irrigating pure air samples with citrus oil
components
Draw-lift’s ZALIMP pump type 335BRychter type washerTwo foil sleeves
Materials
Research methodology
SAMPLING
• taking samples of pure air
Stroehlein Gas Cylinder
Accumulatore
Heat-resistant foil sleeve
Polietylen hose
• irrigating pure air samples with citrus
oil components
Draw-lift’s ZALIMP pump type 335BRychter type washerTwo foil sleeves
• injecting the pollutants: acetone, ethanol,
isopropanol, isoamyl acetate and dillutions of
the basic sample
Materials
Hamilton syringe 500 ml
pure air
Two foil sleeves containing:mixture of air and volatile citrus oil components
&
Research methodology
Schedule of measurements
Research methodology
b. CHROMATOGRAPHIC ANALYSIS in variable temperature conditions
GAS - CHROMATOGRAPH Chromatron GCHF 18.3:
six-permeable tap, sample loop of 5 cm3 capacity,tower 2 metres long with cross-section of 4 mm,packing: Chromosorb W NAW, 60 – 80 mesh, coated with 20% Carbowax 20 M, portative gas: nitrogen, pressure at the inlet 1,2 at, Flame Ionisation Detector, hydrogen pressure 0,4 at, air pressure 0,9 at, detection sensitivity of 30 · 108.
14 defining variables measured [mm] from an invariable basis make a part of a set of data
Defining variables, heights of sequential peaks of a chromatogram [mm]
Number of a standard >10 10-9 9-8 8-7 7-6 6-5 5-4 4-3 3-2 2-1 <1
Sensibility threshold XOdour of a sample X
is a method of evaluating some features of a sample like odour intensity by a group of panelists• 12 students•15 sessions, 10-15 samples during one session
• Basic dilution: 8 cm3 of n-buthanol in 100 cm 3 H2O
• Step of diluting: 2,86
Research methodology
During ANN tests Network Creation Wizard function available in Statistica Neural Network (StatSoft) programme was used.
Multilayer Perceptron and Back Propagation method was applied.
Data set consisted of 14 defining variables (input layer of ANN) and one defined variable (output layer).
a – number of a session of measurements,b – number of a test,i – following number of a studied feature of a pattern,q – total number of patterns in a test.
Irrelative error: proportional share of cases for which difference between sensory and ANN assessment was not graeater than 0,5 in total set of cases.
RESULTS OF THE RESEARCH
Results of the research
Set 1, test 14
Training stages
RMS
Erro
r
0,14
0,16
0,18
0,20
0,22
0,24
0,26
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
TrainingVerification
Set 1, test 14RMS Error - trainig: 0,1686; verification: 0,1552, test: 0,1678
You can conclude that:1. artificial neural networks can properly determine
intensity of air polluted with many compounds,
2. to conduct a training a series of 491 patterns of sensory – chromatographic characteristics of 57 samples evaluated by more than ten people are necessary,
3. it is favorable to remove from the series the estimations of those people whose olfactory sensibility differs considerably from the average,
4. it seems possible to use training series carrying less information of a sample composition for network training.