ABSTRACT Recently there have been advances in automatic monitoring systems for water quality linked to environmental modeling. The difficulty of performing such sensing in natural environments leads to a high probability of faults and... more
ABSTRACT Recently there have been advances in automatic monitoring systems for water quality linked to environmental modeling. The difficulty of performing such sensing in natural environments leads to a high probability of faults and demands for high density data make automated self-test an important requirement. As part of an evaluation of failures in Lab-on-Chip (LOC) Ion Chromatography (IC) and Capillary Electrophoresis (CE) based systems, we passed river water through LOC devices for long periods showing that such system are feasible, but illustrating that failures are likely. Conductivity is the preferred method of detection for IC and CE, but a problem with all such detectors is the need to suppress background conductivity. Chemical suppression is often utilized, but in-situ systems can be simplified using electronic methods. We have developed an approach which directly suppresses baseline current. Detection signals and baseline profiles can be used to diagnose faults, typically manually by an experienced technician. In this paper we discuss automated test strategies for in-situ monitors at both the sensor network and individual instrument level. We also report preliminary investigations into development of automated fault detection using baseline and detection signals.
