Relevance of Drift Components and Unit-to-Unit Variability in the Predictive Maintenance of Low-Cost Electrochemical Sensor Systems in Air Quality Monitoring.

Georgi Tancev

doi:10.3390/s21093298

Abstract

As key components of low-cost sensor systems in air quality monitoring, electrochemical gas sensors have recently received a lot of interest but suffer from unit-to-unit variability and different drift components such as aging and concept drift, depending on the calibration approach. Magnitudes of drift can vary across sensors of the same type, and uniform recalibration intervals might lead to insufficient performance for some sensors. This publication evaluates the opportunity to perform predictive maintenance solely by the use of calibration data, thereby detecting the optimal moment for recalibration and improving recalibration intervals and measurement results. Specifically, the idea is to define confidence regions around the calibration data and to monitor the relative position of incoming sensor signals during operation. The emphasis lies on four algorithms from unsupervised anomaly detection—namely, robust covariance, local outlier factor, one-class support vector machine, and isolation forest. Moreover, the behavior of unit-to-unit variability and various drift components on the performance of the algorithms is discussed by analyzing published field experiments and by performing Monte Carlo simulations based on sensing and aging models. Although unsupervised anomaly detection on calibration data can disclose the reliability of measurement results, simulation results suggest that this does not translate to every sensor system due to unfavorable arrangements of baseline drifts paired with sensitivity drift.

Highlights

Due to the known adverse effects of air pollution, cities consider monitoring relevant air quality indices a necessity [1,2,3]
The output of the four algorithms is consistent and the curves are overlapping, thereby suggesting normally distributed data, as otherwise the Robust covariance (RC) model would deviate from the nonparametric methods; this is true for both sensor (Figure 2a) and reference data (Figure 2b)
The curves of the reference data match with the one of the sensor data, which suggests that the generated anomalies are a result from concept drift and a change in the underlying distribution, i.e., conditions that have not been observed during the calibration phase, as the references would not contain any anomalies otherwise

Summary

Introduction

Due to the known adverse effects of air pollution, cities consider monitoring relevant air quality indices a necessity [1,2,3]. To satisfy the desire for higher spatial and temporal resolution due to differences in individual exposure [5], a lot of research on affordable air quality monitoring devices has been performed [6,7,8,9,10,11,12,13,14,15] Such devices consist of gas and/or particulate matter sensors in the low-cost range, and they are supposed to be connected to the internet of things, forming a network and being part of the smart city vision [16]. Note that Equation (1) is usually inverted to compute the fractions of the target compounds

Objectives

Methods

Results

Conclusion