Data Driven Concept for Sensor Data Adaptation of Electrochemical Sensors for Mobile Air Quality Measurements. Issue 4 (21st February 2020)

• Record Type: Journal Article
• Authors: Esatbeyoglu, Enes; Cassebaum, Oliver; Arras, Florian; Saake, Gunter
• Is Part Of: Journal of the Electrochemical Society. Volume 167:Issue 4(2020)
• Publication Date: 2020-02-21
• Subjects: Electrochemistry -- Periodicals; 541.3705
• DOI: 10.1149/1945-7111/ab74bd
• Languages: English
• ISSNs: 0013-4651

Abstract

Abstract : Electrochemical NO2 sensors are a promising technology for the collection of fine-granular air quality data. Their low acquisition costs, miniaturized size and low power consumption predestine the sensors for use in mobile sensor networks due to their installation in vehicles. However, the sensors are influenced by sensitivities to changes in measurement conditions (e.g. temperature and relative humidity) during data collection. Therefore, we investigated the feasibility of adapting the data of electrochemical NO2 sensors to a reference measuring system using Machine Learning models. For this purpose, we carried out a field experiment with a vehicle to collect data. In addition to three identical electrochemical NO2 sensors, we installed a reference measuring system and measured the NO2 concentration at the same sampling point outside the vehicle. We implemented several adaptation models by using a Multivariate Linear Regression (MLR) and a Support Vector Regression (SVR) to improve the raw sensor signal. Furthermore, we investigated whether an adaptation concept for a single sensor can be applied to identical sensors as well as to other (unknown) measurement runs. The results of this work show the potential of the sensors for air quality assessment, but the accuracy of the reference sensors is only partially achieved.