Pruning of Discrete Dynamic Event Trees Using Density Peaks and Dynamic Time Warping
- Sandia National Laboratories: P.O. Box 5800, Albuquerque, NM 87185 (United States)
Discrete Dynamic Event Trees (DDETs) are used in probabilistic risk assessment (PRA) to determine the risk associated with complex systems, such as nuclear power plants, where accident scenarios need to be realistically modeled. One of the most challenging aspects of the DDET method is that a large number of branches are produced for each initiating event. This can result in data that is difficult to organize and time consuming to analyze. As the number of uncertain parameters increases, so does the complexity of the event tree. An event tree often has nodes that do not contribute to the overall results and predictive capabilities; therefore, we suggest a pre-pruning algorithm that can work to remove these nodes, and consequently, reduce the complexity of the tree. Sandia National Laboratories (SNL) is currently working on a two year dynamic cyber-risk informed project to assess cyber failure in a nuclear power plant. This will be achieved by coupling SNL's dynamic scheduler known as Analysis of Dynamic Accident Progression Trees (ADAPT) with MELCOR, a reactor accident simulator, and a System for Circuit Evaluation and Prediction of Transient Radiation Effects (SCEPTRE), a cyber Emulytics code. This paper will discuss the preliminary results of a feasibility study for using density peaks (DP) clustering with dynamic time warping (DTW) to prune branches that are formed from large DDETs, like the ones that will be produced throughout SNL's project. Papers from the project that are submitted to this conference discuss the following: - The MELCOR model for a Pressurized Water Reactor that uses a digital instrumentation and control system. - Proposed generic cyber exploits for the above reactor. - Modifications to ADAPT to extend the framework for multiple simulators, incorporate importance measures into DDET analysis, as well as sort and reduce DDETs based on inputs to or outputs from the tree. (authors)
- OSTI ID:
- 23042711
- Journal Information:
- Transactions of the American Nuclear Society, Journal Name: Transactions of the American Nuclear Society Vol. 115; ISSN 0003-018X
- Country of Publication:
- United States
- Language:
- English
Similar Records
Dynamic Importance Measures in the ADAPT Framework
Conditional Tree Reduction in the ADAPT Framework
Related Subjects
97 MATHEMATICS AND COMPUTING
ALGORITHMS
DENSITY
FEASIBILITY STUDIES
MODIFICATIONS
NUCLEAR POWER PLANTS
PROBABILISTIC ESTIMATION
PWR TYPE REACTORS
RADIATION EFFECTS
REACTOR ACCIDENTS
REACTOR CONTROL SYSTEMS
REACTOR INSTRUMENTATION
RISK ASSESSMENT
SIMULATORS
TRANSIENTS