Drift compensation of the gas sensor based on self-training and semisupervised learning

Abstract:
The electronic nose is composed of a group of gas sensor arrays. It is often used to detect the composition and concentration of gas. Sensor drift is the main reason affecting its accuracy. In this article, a self-training and semisupervised learning method is proposed. We use the strategy of transductive learning to train the model, which effectively suppresses the drift of the sensor and improves the accuracy of gas classification. It has a certain reference significance in the actual application of gas identification and detection.