Keywords

data quality assessment; fault detection; on-line instrumentation

Location

Session G2: Data Mining for Environmental Sciences (s-DMTES IV)

Start Date

17-6-2014 2:00 PM

End Date

17-6-2014 3:20 PM

Abstract

Compared to traditional grab sampling modern measurement systems enable continuous water quality monitoring of water systems at high frequency. However, in real world applications on-line sensors are still subject to functional, technical and operational constraints. Challenges thus remain associated with the automation of data collection and especially data validation to ensure proper use and interpretation of the data and avoid the danger of building data graveyards. Poor quality data could drastically affect the results of their application, e.g. water quality models for river basin management, model-based control, WWTP control rules, decision making, etc. For practical fault detection purposes, in this paper, a data-driven tool that attempts to extract useful information from the time series of multiple measurement signals, in the absence of exact process knowledge, is presented. The proposed tools are successfully tested on on-line water quality time series from different applications including sewers, wastewater treatment plants and receiving waters.

Share

COinS
 
Jun 17th, 2:00 PM Jun 17th, 3:20 PM

Automated data quality assessment: Dealing with faulty on-line water quality sensors

Session G2: Data Mining for Environmental Sciences (s-DMTES IV)

Compared to traditional grab sampling modern measurement systems enable continuous water quality monitoring of water systems at high frequency. However, in real world applications on-line sensors are still subject to functional, technical and operational constraints. Challenges thus remain associated with the automation of data collection and especially data validation to ensure proper use and interpretation of the data and avoid the danger of building data graveyards. Poor quality data could drastically affect the results of their application, e.g. water quality models for river basin management, model-based control, WWTP control rules, decision making, etc. For practical fault detection purposes, in this paper, a data-driven tool that attempts to extract useful information from the time series of multiple measurement signals, in the absence of exact process knowledge, is presented. The proposed tools are successfully tested on on-line water quality time series from different applications including sewers, wastewater treatment plants and receiving waters.