As artificial intelligent (AI) technology has been dramatically developed, various industries have been challenged to apply it. In a view of nuclear power plants (NPP), it seems that AI technology applies to NPPs at the last because NPPs are required the most stringent level of regulatory guideline for safety. To overcome it, AI technology should be applied incrementally into the NPPs rather than all at once. According to the unintended shutdown records during startup and shutdown operation from 1997 to 2017 in Korea, it is reported that human errors accounts for 40% of the total. This is because operators feel heavy burden to monitor hundreds of parameters for a long time of operating time. Also, there are lots of startup and shutdown operating history that can be used for correcting the data from the NPP simulator. Therefore, this work proposes a framework to develop AI automatic operating system for startup and shutdown operations of NPPs. Operating procedures of startup and shutdown operations are categorized. In addition, AI technologies will be introduced to find out the most suitable learning algorithm. It is expected that economic loss from human error during startup and shutdown operation will be reduced as AI system developed.
[1]
Foudil Abdessemed,et al.
Simulation of the Navigation of a Mobile Robot by the QLearning using Artificial Neuron Networks
,
2009,
CIIA.
[2]
Poong Hyun Seong,et al.
ADVANCED MMIS TOWARD SUBSTANTIAL REDUCTION IN HUMAN ERRORS IN NPPS
,
2013
.
[3]
Shie Mannor,et al.
Bayesian Reinforcement Learning
,
2012,
Reinforcement Learning.
[4]
Alex Graves,et al.
Playing Atari with Deep Reinforcement Learning
,
2013,
ArXiv.
[5]
Poong Hyun Seong,et al.
Development of automated operating procedure system using fuzzy colored petri nets for nuclear power plants
,
2004
.