NRC Examines Possible Role of Artificial Intelligence in Commercial Nuclear Power Operations | Morgan Lewis – Up & Atom

With the proliferation of artificial intelligence (AI) and machine learning tools in various products and industries, the NRC has begun to investigate what role these technologies can play in commercial nuclear power operations. On April 21, as part of its study, the NRC’s Bureau of Nuclear Regulatory Research requested public comments on the role of these technologies “in the various phases of nuclear power generation operational experience and asset management.” The NRC is seeking feedback on “the state of the art, benefits and future trends in relation to [these technologies’] Computational tools and techniques for predictive reliability and predictive safety assessment in the commercial nuclear power industry. “These technologies are” emerging analytical tools that, when used properly, offer promising opportunities to improve reactor safety while still offering economic savings. “Comments are due until May 21, 2021.

The NRC intends to use the comments to improve its understanding of the benefits of AI and machine learning and the “potential pitfalls and challenges associated with using them.”

The NRC asked for comments on the following questions:

  1. What is the state of development or use of AI / machine learning tools in the commercial nuclear power industry to improve aspects of the design, operation, maintenance or decommissioning of nuclear power plants? What tools are used or developed? When should the tools currently under development be used?
  2. Which areas of the commercial operation and management of nuclear reactors will benefit most and least from implementing AI / machine learning? Possible examples include inspection support, incident response, power generation, cybersecurity, predictive maintenance, safety / risk assessment, monitoring of system and component performance, operational / maintenance efficiency, and shutdown management.
  3. What are the potential benefits to commercial nuclear power operations from the inclusion of AI / machine learning in terms of (a) design or operational automation, (b) preventive maintenance trends, and (c) improved productivity of reactor operations staff?
  4. Which AI / machine learning methods are currently or in the near future used in the commercial management and operation of nuclear power plants? Examples of possible AI / machine learning methods include, but are not limited to, artificial neural networks, decision trees, random forests, support vector machines, clustering algorithms, algorithms for dimension reduction, data mining and content analysis tools, Gaussian processes, Bayesian methods, processing natural language and image digitization.
  5. What are the pros or cons of having a top-down, high-level strategic objective for developing and implementing AI / machine learning in a wide range of general-purpose applications versus a case-by-case, ad hoc approach?
  6. What stage of technology adoption is the commercial nuclear power industry currently experiencing in terms of AI / machine learning and why? The current model of technology introduction characterizes phases in categories such as the innovation phase, the early adopter phase, the early majority phase, the late majority phase and the Laggard phase.
  7. What challenges arise from weighing the costs associated with the development and use of AI / machine learning tools against the operational and technical advantages of the plant in integrating AI / machine learning into operational decision-making and workflow management?
  8. What is the general level of AI / machine learning in the commercial nuclear power industry (e.g., expert, savvy / skilled, or novice)?
  9. How will AI / machine learning affect the commercial nuclear power industry in terms of efficiency, cost and competitive position compared to other power generation sources?
  10. Does AI / machine learning have the potential to improve the efficiency and / or effectiveness of nuclear oversight or otherwise affect the regulatory costs associated with safety oversight? If so, in what way?
  11. AI / machine learning usually requires the creation, transmission and evaluation of very large amounts of data. What are some data security concerns related to proprietary nuclear power plant operating experience and design information that may be stored on remote external networks?

The NRC is in the early stages of its review and the agency does not promise to use the information it collects for any formal regulatory action. Morgan Lewis will continue to follow the NRC’s regulatory initiatives.

[View source.]

Comments are closed.