Assessing the Relocation Robustness of on Field Calibrations for Air Quality Monitoring Devices
The adoption of on field calibration for pervasive air quality monitors, is increasing significantly in the last few years. The sensors data, recorded on the field, together with co-located reference analyzers data, allow to build a knowledge base that is more representative of the real world conditions and thus more effective. However, on field calibration precision may fade in time due to change in operative conditions, due to different drivers. Among these, relocation is deemed among the most relevant. In this work, for the first time, we attempt to assess the robustness of this approach to relocation of the sensor nodes. We try to evaluate the impact on performance of the so called locality issue by measuring the changes in the performance indicators, when a chemical multisensory system operates in a location that differs from the one in which it was on field calibrated. To this purposes, a nonlinear multivariate approach with Neural Networks (NN) and a suitable dataset, provided by NILU (the Norwegian Institute for Air Quality), have been used. The preliminary results show a greater influence of seasonal forcers distribution with respect to the relocation issues.
KeywordsMobile chemical multisensory devices Machine learning Distributed air quality monitors Calibration methods