The model of odour intensity analyzer in a form of two layers of neurons. was proposed. The first layer consists of specific receptors of odorants, which are randomly distributed. The odorant influence on odour intensity of mixtures is depends on the number of suitable receptors and number of their connections with the second layer neurons (the degree of divergence). The second layer neuron is stimulated with the same power after receiving the signal from any receptor or receptors. Il was assumed that the odour intensity is directly proportional to the number of stimulated neurons in the second layer. The investigated model contained 1000 neurons in the first layer and 1456 neurons in the second layer. The first layer contained receptors A and B (RA and RB) in quantities 15% RA, IS% RB and 30% RA, 10% RB. Receptors A characterized the divergence I :9, receptors B - 1 :25. The stimulus were mixtures containing 5-100 molecules A and B per hundred (x , = 0.0, 0.2, 0.5, 0.8 and 1.0). The general similarity of • the model reaction with that of the human smell reaction was found.

Field measurements of the near-ground odorant concentrations were made in surrounding of the mechanical sewage treatment plant with separated digestive chamber. The concentrations downwind the pollutants plume in several different meteorological situations were the basis for the estimations by a process of trial and error. The estimations encompassed the total emission of odorants and emission rates related to the volume of the purified sewage, to the amount of the removed pollutants and to the area of the unorganized emission sources during the autumn. Parallel research concerning other seasons and different surface sources of odorants are planned.

Searching for dependance of odour intensity on degree of dillution of basic samples with pure air, samples aromatised with citrus oil and containing difTerent amounts of acetone, ethanol, isopropanol, isoamyl acetate were studied. 192 individual odour intensity assesments of 16 samples were collected. Dependence of odour intensity on acetone concentration determined with chromatographic analysis (110-16500 mg/m3) was defined on the basis of 311 individual odour intensity assessments of 24 samples. The measurements' results were elaborated without initial assumptions concerning character of a psychophisical equation and without initial selection of empirical data. Automatic Networks Designer (Statistica Neural Network, StatSoft) was used. Dependence of odour intensity on concentration was described with logistic function resembling Stevens law within a small concentrations range and Weber-Fechner law within a high concentrations range.

The possibility of applying gas chromatograph to air samples discrimination in regard to odour intensity and hedonic quality was examined. The air samples were arornatised with lemon oil and four admixtures. fourteen distinctive points of a chromatogram were appointed and the distances from the points lo a set basis were measured. The set of h 1-h 14 parameters (inputs) and varied individual sensory estimations of I and H (outputs) was used as a training data set for NN. Possibility of discrimination of the odour quality of the samples situated close to the threshold of odour quality difference detectability was confirmed on the level of approximately one.

Assessments of odour nuisance are made using one of the psychophysical laws: logarithmic Weber- Fechner law or power law by Stevens. The object of the work was to determine whether one of the laws exhibits advantage over the other. To evaluate the two laws series of odorimetric measurements of aqueous solutions of acetic acid were made. Values of Pearson' correlation coefficient were calculated for relationship of the odour intensity (I) versus logarithm of the acetic acid concentration in air (Iog., C [mg/rrr'I) as well as for the relationship of logarithm of the odour intensity (log,0/) versus logarithm of the acetic acid concentration in air (log., C [rng/nr'[). Both laws are fulfilled equally well in a range of investigated concentrations of acetic acid.

550 individual odour intensity (S) assessments of 62 samples of air polluted with 2-propanol were collected (concentration c = 46-2 1000 mg/m3). The data were used to establish odour detection threshold of 2-propanol by extrapolation to S = 0 in the coordinate system of S-log c. The obtained values were compared with triangular tests results and data published in the literature. A reason for apparent positive deviations from Weber-Fechner law within a range of low odour intensities was indicated,

Evaluations of odour intensity of the air polluted with cyclohexane (A), cyclohexanone (B) and cyclohexane with admixtures of cyclohexanone (mixtures MI, M2 and M3; x8 = 0.14; 0.05 and 0.02 respectively) were made. Empirical parameters of psychophysical equations (Weber-Fechner' and Stevens') were determined. Total concentrations of the mixtures were within the range C = 1500-3500 ppm. The interaction constant av= 115-132° was estimated for the mixtures. Verification of known models of odour interaction led to the conclusion that possible application of the models for prediction of odour intensity of the air polluted with mixtures of cyclohexane and hexane is limited.