A chemiresistive sensor array based on polyaniline nanocomposites and machine learning classification.

Area of Science:

• Materials Science
• Chemical Sensing
• Machine Learning

Background:

• Accurate detection of gases like ammonia (NH3), nitrogen dioxide (NO2), carbon oxides (CO2, CO), acetone, and toluene is crucial for environmental monitoring and industrial safety.
• Polyaniline nanocomposites offer promising properties for developing sensitive and selective gas sensors.

Purpose of the Study:

• To investigate the selective detection of ammonia (NH3), nitrogen dioxide (NO2), carbon oxides (CO2 and CO), acetone, and toluene using a gas sensor array.
• To evaluate the performance of various machine learning algorithms for accurate gas classification.

Main Methods:

• Fabrication of a seven-sensor array with different conductive polyaniline nanocomposite sensing layers.
• Application of dimensionality reduction techniques, including principal component analysis (PCA) and linear discriminant analysis (LDA).
• Comparison of five classification methods: k-nearest neighbors, support vector machine, random forest, decision tree classifier, and Gaussian process classification (GPC).

Main Results:

• Principal component analysis effectively reduced data dimensionality for improved classification.
• Gaussian process classification (GPC) demonstrated superior performance among the tested algorithms.
• The GPC model, trained on PCA-extracted features, achieved a highly accurate classification rate of 99% for six different gases.

Conclusions:

• The developed polyaniline nanocomposite-based gas sensor array, coupled with machine learning, offers a highly effective solution for selective gas detection.
• Gaussian process classification, enhanced by PCA, represents a robust method for analyzing sensor array data and accurately identifying target gases.
• This approach holds significant potential for real-time monitoring applications in environmental and safety-critical fields.