Nuclear

How Generative AI could help nuclear plant operators

New research from Argonne National Laboratory shows how cutting-edge AI could change the way nuclear plant workers learn to detect and solve problems.
How Generative AI could help nuclear plant operators
(Source: Medium.)

Generative AI models could enhance how nuclear plant operators handle complex diagnostic information, according to new research from the U.S. Department of Energy’s (DOE) Argonne National Laboratory.

The goal is to not only detect faults, but improve decision-making by presenting diagnostic information in clear, understandable terms that detail what is wrong, why it is wrong and how it can be addressed.

Argonne engineers combined three elements: an Argonne diagnostic tool called PRO-AID, a symbolic engine and an LLM in their research. The diagnostic tool uses facility data and physics-based models to identify faults. The symbolic engine acts as an intermediary between PRO-AID and the LLM. It creates a structured representation of the fault reasoning process and constrains the output space for the LLM, which acts to eliminate hallucinations. Then, the LLM explains these faults in a straightforward manner for the operators.

“The system has the potential to enhance the training of our nuclear workforce and streamline operations and maintenance tasks,” says Rick Vilim, manager of the Plant Analysis and Control and Sensors department at Argonne.

PRO-AID works by comparing real-time data from the plant to expected normal behaviors, Argonne said. When there’s a mismatch, it indicates a fault. This process involves using models that simulate the plant’s components and how they should normally behave. If something doesn’t match, there’s a problem, and PRO-AID provides a probabilistic distribution of faults based on these mismatches.

A key challenge with LLMs is ensuring they provide accurate information, Argonne engineers said. The authors address this by designing a symbolic engine to manage the information the LLM uses, ensuring it only provides explanations based on the data and models.

The LLM is used to explain the results from PRO-AID. It takes complex technical data and translates it into easy-to-understand language. This helps operators understand the cause of the fault and the reasoning behind the diagnosis. Additionally, using natural language, the operators can use the LLM to inquire arbitrarily about the system and sensor measurements.

The system was tested at Argonne’s Mechanisms Engineering Test Loop Facility (METL), where small- and medium-sized components are tested for use in advanced, sodium-cooled nuclear reactors. The system diagnosed a faulty sensor and explained the issue to the operators. Argonne engineers say this demonstrated that combining a diagnostic tool with an LLM can effectively provide understandable and trustworthy explanations for faults in complex systems.

See Argonne National Laboratory’s research in its new paper here.

Tags