Algorithm for the Signal Validation in the Emergency Situation Using Unsupervised Learning Methods
| Autor: | Gyeongmin Yoon, Younhee Choi, Jonghyun Kim |
|---|---|
| Rok vydání: | 2021 |
| Předmět: | |
| Zdroj: | Lecture Notes in Networks and Systems ISBN: 9783030806231 AHFE (13) |
| DOI: | 10.1007/978-3-030-80624-8_33 |
| Popis: | The safe operation of nuclear power plants (NPPs) requires valid and correct signals. Faulty signals and sensors may deteriorate the performance of both control systems and operators in emergency situations, as learned from past accidents such as Three Mile Island and Fukushima Daiichi NPPs. Moreover, there is increasing interest relating to autonomous and automatic controls because successful control largely relies on input signals’ integrity and reliability. This study proposes an algorithm for signal validation in emergency situations at NPPs using unsupervised learning methods. The algorithm employs unsupervised learning based on a combination of Variational Auto-Encoder (VAE) and Long Short-Term Memory (LSTM). The proposed algorithm is also validated using the Compact Nuclear Simulator (CNS). The results show that the proposed algorithm can successfully detect more than 97.6% of signal failures in emergency situations. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|