Quantitative odour modelling using electronic nose information

The odour samples from piggeries were analyzed simultaneously using the AromaScan “electronic nose” and the industry standard technique of olfactometry. The resulting sensor data has been used to train a feedforward backpropagation artificial neural network to correlate the responses to the odour units. The neural network has also been trained by applying principal component analysis on the sensor data with a view to reducing the dimensionality of the problem and computational time taken for training. Principal component analysis finds the projections of maximum variance. The number of features needed for effective data representation is reduced by discarding linear dependencies that exist among the variables