Integrated Decision Making for Nuclear Power Plant Maintenance using Deep Reinforcement LearningSpangler, Ryan (2024) Integrated Decision Making for Nuclear Power Plant Maintenance using Deep Reinforcement Learning. Doctoral Dissertation, University of Pittsburgh. (Unpublished) This is the latest version of this item.
AbstractThe goal of this research is to improve nuclear power plant operations and maintenance decision-making by integrating condition monitoring and deep reinforcement learning to reduce overall lifetime costs. Although nuclear power is one of the most reliable and available forms of energy production, it remains costly. For nuclear power to be a viable and competitive source of energy generation, these costs must be reduced. Currently, operation and maintenance (O\&M) relies on overly conservative time-based inspections without using knowledge of the current condition of the plant. This causes unexpected failures, unnecessary inspections, and conservative repairs, the combination of which results in high O\&M costs. Further complicating O\&M decision making is the strict operating requirements and constraints that are unique to the nuclear industry. Challenges such as refueling outage schedules, costly unplanned shutdowns, limited supply chain manufacturers, long lead times for high-value assets, and the inability to quickly shutdown and restart production make nuclear O\&M difficult. As part of this research, we have developed an asset management tool using deep reinforcement learning (DRL) that is capable of reducing lifetime maintenance and repair costs through improved decision making. Using the latest advancements in condition monitoring, supply chain analytics, and deep reinforcement learning, we have created a predictive maintenance tool that optimizes the asset management of integrated nuclear systems. By providing an improved asset management tool, we can reduce overall lifetime costs by helping nuclear operators make synchronized and consistent decisions for several components over the span of multiple outages. Share
Details
Available Versions of this Item
MetricsMonthly Views for the past 3 yearsPlum AnalyticsActions (login required)
|
![[feed]](/images/feed-icon-32x32.png){kind=icon}