Data requirements for an anomaly detector in an automated safeguards system using neural networks
- Los Alamos National Lab., NM (United States). Safeguards Systems Group
- Westinghouse Idaho Nuclear Co., Idaho Falls, ID (United States)
An automated safeguards system must be able to detect and identify anomalous events in a near-realtime manner. This approach to anomaly detection is based on the demonstrated ability of neural networks to model complex, nonlinear, real-time processes. By modeling the normal behavior of processes, one can predict how a system should behave and, thereby, modeling the normal behavior of processes, one can predict how a system should behave and, thereby, detect when an abnormal state or event occurs. In this paper, the authors explore the computational intensity of training neural networks, and they discuss the issues involved in gathering and pre-processing the safeguards data necessary to train a neural network for anomaly detection. They explore data requirements for training neural networks and evaluate how different features of the training data affect the training and operation of the networks. They use actual process data to train the previous 3-tank model and compare the results to those achieved using simulated safeguards data. The data represented tank volumes and sensor states from M-cell processing tanks in the Idaho Chemical Processing Plant. Comparisons are made on the basis of required training times in addition to correctness of prediction.
- OSTI ID:
- 61301
- Report Number(s):
- CONF-930749--
- Country of Publication:
- United States
- Language:
- English
Similar Records
Anomaly detection in an automated safeguards system using neural networks
Anomaly detection in an automated safeguards system using neural networks