The Nuclear Science and Engineering (NSE) Division is developing artificial intelligence (AI) and machine learning (ML) methods with the objective of improving the economic competitiveness of nuclear power. Applications to the existing fleet are underway in parallel with development of methods for advanced reactors. Advanced reactors present an opportunity for seamless integration of AI/ML methods as many of the designs are at the conceptual design stage.
AI/ML is being used to address the challenge of autonomous operation in an environment where the reactor is part of a dynamic energy network. In existing commercial reactors automation has a very limited role with the plant designed to produce base load power. In the vision of the future, the human operator fitted with AI/ML technology operates in a state of enhanced situational awareness acting in an oversight role while the complex demands of monitoring and control are delegated to a machine. The greatest potential for enabling this vision exists for the advanced reactor. There are, however, elements also of interest to utilities with reactors that are to undergo license renewal.
The transformative potential of AI/ML resides in the algorithms, which can find patterns and correlations in data and can mimic how humans process data to reason and make judgments in performing goal-oriented tasks. Machine algorithms have the potential assume the execution of labor-intensive goal-driven tasks currently performed by humans. In the envisioned application the human cognitive role is reshaped from performing data-intensive low-level tasks to higher-level oversight. The primary benefit is a significant reduction in staffing. Independent and diverse lines of safety are maintained with the inherent safety features of the advanced reactor and the plant protection system of the existing fleet.