The idea is that several complementary information sources (mainly sensors in addition to process knowledge) can be used to capture all the information needed and to draw more robust conclusions that cannot be obtained from any of the sources alone.

One of the challenges is to determine how different measurements of varying qualities are to be combined to give accurate and reliable results. If not properly done a bad measurement can sometimes deteriorate an already good estimate, while in other situations the same measurement may be the only good source of information available. It is thus necessary to take into account which and when the different measurements are relevant and how for instance contradictory information shall be weighted.

For this we use what is available of physical understanding and models of the observed system and the measuring systems, empirical statistical models for the accuracy and reliability of the individual measurements, and multivariate analysis and modeling techniques to combine the information.

Another big challenge is to achieve reliable deviation detection (or fault detection) in the harsh and demanding offshore environment, with a minimum of false alarms. This challenge has lots of resemblance with outdoor surveillance systems, and SINTEF has extensive experience in developing robust outdoor surveillance systems collaborating since 1992 with Detec A/S in developing their automatic video surveillance systems.

This project is one of the research activities in in the ongoing project Robotics for oil & gas platforms. 

The screenshot is from the process viewing software we use for robot control and process surveillance. What we see is an example of how the operator can be warned that there is an abnormal situation detected at several different measuring positions in the process equipment.