Classification with quantification for air quality monitoring

Abstract

In this paper, a fuzzy classification with quantification algorithm is proposed for solving the air quality monitoring problem using e-noses. When e-noses are used in dynamic outdoor environment, the performance suffers from noise, signal drift and fast-changing natural environment. The question is, how to develop a prediction model capable of detecting as well as quantifying gases effectively and efficiently? The current research work has focused either on detection or quantification of sensor response without taking into account of dynamic factors. In this paper, we propose a new model, namely, Fuzzy Classification with Quantification Model (FCQM) to cope with the above mentioned challenges. To evaluate our model, we conducted extensive experiments on a publicly available datasets generated over a three-year period, and the results demonstrate its superiority over other baseline methods. To our knowledge, gas type detection together with quantification is an unsolved challenge. Our paper provides the first solution for this kind.