Enhancing the Discrimination Ability of a Gas Sensor Array Based on a Novel Feature Selection and Fusion Framework.

Area of Science:

• * Chemical sensing and electronic noses
• * Signal processing and machine learning

Background:

• * Gas sensor arrays are crucial for odor identification but face challenges in discrimination ability.
• * Effective feature selection and fusion are vital for improving the performance of gas sensor systems.

Purpose of the Study:

• * To propose a novel feature selection and fusion framework for enhancing gas sensor array discrimination.
• * To improve odor identification accuracy using advanced signal processing techniques.

Main Methods:

• * An efficient feature selection method based on separability and dissimilarity was developed.
• * K-nearest neighbor (KNN) classifier was used to determine optimal feature subset dimensions.
• * A classification dominance feature fusion strategy was implemented.

Main Results:

• * Achieved 97.5% recognition rate on Database I and 80.11% on Database II using KNN (k=1) with correlation distance.
• * Demonstrated the framework's superiority in signal feature representation.
• * The method effectively selects features and fuses diverse sensor signal characteristics.

Conclusions:

• * The proposed framework significantly enhances the discrimination ability of gas sensors for odor identification.
• * The feature selection and fusion strategy effectively represents signal features and suppresses drift effects.