Abstract A new dynamic gas classification model was developed to achieve a reliable on-line discrimination at very fast response times. The aim was to be able to follow rapid changes in gas compositions using an electronic nose in consumer applications. The electronic nose is based on a microarray especially designed for production at very low costs. This is essential for application in mass products. Common classification methods used for signal evaluation of electronic noses such as Linear Discriminant Analysis (LDA), Neural Networks (NN) or Soft Independent Modelling of Class Analogy (SIMCA) fail to detect non-stationary gas mixtures. The new model, however, combines classification of steady states with transient evaluation via time series analysis. Rapid signal transients are detected by appropriate digital filters, steady state signals are classified by the above mentioned standard methods. The simplicity of the algorithm model allows implementation in low-cost electronic units, containing micro controllers with very limited memory capacity. To give an example, the automatic control of the ventilation flap of automobiles was investigated. Intermediate streams of bad air could be detected within 1–2 s. The error of pollutants detection was reduced from 25%, applying static classification only, to 10% for the new dynamic model.
[1]
M. Marth,et al.
Two improvements of early transition detection
,
1999
.
[2]
Les Thede,et al.
Analog and digital filter design using C
,
1996
.
[3]
Timothy Masters,et al.
Advanced algorithms for neural networks: a C++ sourcebook
,
1995
.
[4]
Svante Wold,et al.
Pattern recognition by means of disjoint principal components models
,
1976,
Pattern Recognit..
[5]
J. Goschnick,et al.
Multisensor microsystem for contaminants in air
,
1996
.
[6]
B. Kowalski,et al.
Partial least-squares regression: a tutorial
,
1986
.
[7]
M. Madou,et al.
Chemical Sensing With Solid State Devices
,
1989
.
[8]
Josef Honerkamp,et al.
Early Transition Detection - A Dynamic Extension to Common Classification Methods
,
1998
.
[9]
P. Hauptmann.
Sensors: A Comprehensive Survey
,
1996
.
[10]
D. Gonzalez-Arjona,et al.
Adaptation of linear discriminant analysis to second level-pattern recognition classification
,
1998
.
[11]
R. Ogden,et al.
Essential Wavelets for Statistical Applications and Data Analysis
,
1996
.