Sensor Circuits - The Electronic Nose Artificial Olfaction Technology - page 193

Fig. 7.21 Digital encoder

descriptive parameters from the sensory array response, and preparing the feature
vector for further processing. Improvements in the performance of the available
data analysis techniques are an important topic of research on electronic nose
development. Figure 7.22 represents the schematic process presentation in data
analysis techniques.
Processing may do the following things.

• Change the level or value of the signal (e.g., voltage level).
• Change the signal from one form to another (e.g., current to pneumatic).
• Change the operating characteristic with respect to time.
• Convert analog and digital signals from one to the other.

Preprocessing Techniques

The raw data were preprocessed to improve the quality of the input data. Data
preprocessing techniques are designed to provide modification of existing raw data
in order to achieve more quality supply in input variable.
Prior to any data analysis, it is usual to carry out preprocessing of the data. The
main aims of this stage are:

1. To reduce the amount of data those are not related to the study.
2. To increase sufficient information within the data to get the desired purpose.
3. To remove the information in, or transform the data to, a form suitable for
further analysis.

Some examples of signal processing algorithms used are shown in Table 7.2
[ 2 ]. (i = sensor,j= odor, a = odor a, b = reference odor b, r = population
standard deviation, X=average value, N = the number of feature vectors in the
feature-set with i component to each vector).