On the Use of Linear Regression for the Assessment of Stability in Noise Monitoring Networks: A Practical Example
Invited paper
National Physical Laboratory
Wednesday 3 june, 2015, 11:20 - 11:40
0.8 Rome (118)
Abstract:
As modern cities grow and urban environments become increasingly complex the
need for creation of robust systems which can be used for monitoring various
environmental parameters including noise becomes more and more profound. Such
systems enable the extraction of useful information about a city’s environment
which can then be used to inform the public. Technological advances have made
possible the creation of systems using cheap components like mems microphones.
However the nature of their components as well as the network’s installation
makes them vulnerable to a great number of destructive factors like extreme
weather conditions and aging of their elements. Ensuring good operation is
vital for the reliability of the recorded data. Hence the creation of
monitoring systems aiming to evaluate data quality is crucial. Over the years
various methods have been developed towards achieving this goal. In this paper
we propose the use of statistical parameters, spatial and temporal
correlations between closely located receptors as well as the possible
utilization of available information on events that can trigger significant
sound levels for the evaluation of a network’s stability in terms of
sensitivity of the receptors and the identification of noise patterns which
can indicate data degradation due to system failures or significant drifts
from calibration. The proposed methods were developed based on datasets
recorded at Katendrecht, Rotterdam between December 2013 and January 2014.