Abstract
A nuclear power plant is instrumented with a variety of sensors that monitor its variables to estimate the state and initiate safety actions, if needed. We address the problem of estimating drifts or errors in sensor measurements due to factors such as calibration changes. We propose an information fusion method that uses measurements from other sensors to generate an estimate of a sensor measurement, and its difference from an actual measurement provides an error estimate. We present two fusers based on the ensemble of trees and support vector machine that are trained using sensor measurements collected at an emulated test loop of a pressurized water reactor under no-drift conditions. We present error estimates for a differential pressure sensor of the heat exchanger of the primary coolant system, under twenty controlled scenarios using the test loop. Both positive and negative errors are captured by both methods in scenarios involving calibration drifts, blocking, air gap and electromagnetic interference. The root mean square errors of estimated drifts are typically within 2% percent of the maximum.
Original language | English |
---|---|
Title of host publication | 2019 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS/MIC 2019 |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
ISBN (Electronic) | 9781728141640 |
DOIs | |
State | Published - Oct 2019 |
Event | 2019 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS/MIC 2019 - Manchester, United Kingdom Duration: Oct 26 2019 → Nov 2 2019 |
Publication series
Name | 2019 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS/MIC 2019 |
---|
Conference
Conference | 2019 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS/MIC 2019 |
---|---|
Country/Territory | United Kingdom |
City | Manchester |
Period | 10/26/19 → 11/2/19 |
Funding
Notice: This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE).
Keywords
- Power reactor
- coolant system
- machine learning
- multiple sensor fusion
- sensor drift