ML22221A242
| ML22221A242 | |
| Person / Time | |
|---|---|
| Issue date: | 07/30/2022 |
| From: | Dennis M, Hathaway T Office of Nuclear Regulatory Research |
| To: | |
| Kenneth Kolaczyk, NRR/DRO/IRAB, 58577389 | |
| Shared Package | |
| ML22221A240 | List: |
| References | |
| Download: ML22221A242 (31) | |
Text
Draft Artificial Intelligence Strategic Plan Fiscal Years 2023-2027 Operating Reactors KM Session July 30, 2022 Matt Dennis Reactor Systems Engineer (Data Scientist)
Office of Nuclear Regulatory Research Dr. Trey Hathaway Reactor Systems Engineer (Code Development)
Office of Nuclear Regulatory Research
Nuclear Regulatory Commission Artificial Intelligence Strategic Plan Background, Purpose and Preparation Increased Industry Interest Regulatory Readiness Common Understanding of the Levels of Autonomy AI Strategic Plan
Developing a Consistent Understanding
- Artificial Intelligence
- A machine-based system that can go beyond defined results and scenarios and has the ability to emulate human-like perception, cognition, planning, learning, communication, or physical action. For a given set of human-defined objectives, AI can make predictions, recommendations, or decisions influencing real or virtual environments. These systems use machine-and human-based inputs to perceive real and virtual environments, abstract such perceptions into models through analysis in an automated manner, and use model inference to formulate options for information or action.1
- An application of artificial intelligence that is characterized by providing systems the ability to automatically learn and improve on the basis of data or experience, without being explicitly programmed.1 1National Defense Authorization Act (2021)
REGULATORY PURPOSE
- NRC recognizes a need to use technology innovation for regulatory enhancements as part of its effort to become a modern, risk-informed regulator1
- The nuclear industry is researching and using artificial intelligence (AI) applications; therefore, the NRC must be prepared to review and evaluate the use of AI in NRC-regulated activities
- The AI Strategic Plan presents the vision and goals for the NRC to cultivate an AI-proficient workforce, keep pace with AI technological innovations, and ensure the safe and secure use of AI in NRC-regulated activities 1 The Dynamic Futures for NRC Mission Areas, (ML19022A178)
PREPARATORY ACTIVITIES
NRC AI Research Projects Past and Present
- Watson Content Analytics for Operational Experience Data
- ADAMS NLP Resource Estimation Tool
- Operating Experience Event Classification
- Named Entity Recognition for Regulatory References
- Regulatory Viability of Digital Twins
Nuclear Industry AI Survey
- In April 2021, NRC issued an FRN soliciting feedback on the nuclear industrys AI readiness and applications
- FRN responses indicated 1.Benefits are great, but cost of developing and implementing AI is a challenge 2.Concerns about data security, including cyber intrusions, proprietary information leakage, and loss of export control 3.Developers are particularly interested in using NLP, neural networks, and clustering algorithms
Three workshops held in 2021 and archived on the NRC website, Nuclepedia, and ML21348A637 Industry expressed interest for the NRC to develop a generic set of AI design criteria in a Regulatory Guide for LWR and non-LWR applications As AI applications rely on access to high quality data, agreement needs to be reached on a centralized entity empowered to aggregate and share data Industry AI application deployment in 1-2 years, and AI applications for NRC regulatory review in 3-5 years FOUNDATIONAL KNOWLEDGE COLLABORATION OPPORTUNITIES DATA SHARING CURRENT PROJECTS FUTURE ACTIVITIES Workshops Objectives and Findings
External Awareness Commenced interactions with other federal partners to explore AI evaluation approaches, lessons learned, and collaboration opportunities Similarly engaged with international counterparts interested in AI for nuclear Maintaining continual engagement on state-of-the-art research Enhancing and leveraging existing MOUs International Counterparts Leveraging MOUs New Data Science and AI Addendum Operating Experience and Data Analytics Other Federal Agencies Research Organizations and Academia
Nuclear Industry AI Applications
- Department of Energy and Industry Project Categories
- Increasing existing NPP economic efficiency
- Plant condition monitoring
- Process improvement and cost reduction
- Plant automation
- Sensor-enabled degradation assessment and wireless security
- Operating Reactor Projects
- CAP Analyzer for informed inspections
- Palo Verde Process Automation
- Core design optimization
- Advanced Reactor Projects
- Digital twins using deep neural networks
- Autonomous operation in remote locations
NRC AI Challenges Future Regulatory Guidance and Decisionmaking Development Differentiating AI Usage for Reactor Design Versus Autonomous Control Explainable AI and Trustworthy AI - Reliability and Assurance Internal AI Budget Predicated on Emergent Industry Applications Current Workforce Training Traceable and Auditable Evaluation Methodologies Internal Challenges: Automating Internal Agency Business Processes External Challenges: Understanding Licensee and Applicant AI Usage
Notional AI and Autonomy Levels in Commercial Nuclear Activities Level Notional AI and Autonomy Levels Potential Uses of AI and Autonomy in Commercial Nuclear Activities Level 1 Insight Human decisionmaking assisted by a machine AI integration in systems is used for optimization, operational guidance, or business process automation that would not affect plant safety/security and control Level 2 Collaboration Human decisionmaking augmented by a machine AI integration in systems where algorithms make recommendations that could affect plant safety/security and control are vetted and carried out by a human decisionmaker Level 3 Operation Machine decisionmaking supervised by a human AI and autonomy integration in systems where algorithms make decisions and conduct operations with human oversight that could affect plant safety/security and control Level 4 Fully Autonomous Machine decisionmaking with no human intervention Fully autonomous AI in systems where the algorithm is responsible for operation, control, and intelligent adaptation without reliance on human intervention or oversight that could affect plant safety/security and control Common Understanding of the Level Key for Regulatory Readiness Human Involvement Machine Independence
AI STRATEGIC PLAN SCOPE
The AI Strategic Plan (ML22175A206) consists of five strategic goals:
- Goal 1: Ensure NRC Readiness for Regulatory Decisionmaking
- Goal 2: Establish an Organizational Framework to Review AI Applications
- Goal 3: Strengthen and Expand AI Partnerships
- Goal 4: Cultivate an AI-Proficient Workforce
- Goal 5: Pursue Use Cases to Build an AI Foundation Across the NRC ARTIFICIAL INTELLIGENCE STRATEGIC PLAN OVERVIEW AI Strategic Plan Development
- NRC AI Team: Engaged interdisciplinary team of AI subject matter experts across the agency
- Leveraged insights from 2021 Data Science and AI Regulatory Applications Workshops (ML21348A637)
AI Strategic Plan Near-Term FY23 Outcomes
- Establish Data Science Training and Qualification Plan
- Establish AI Steering Committee (AISC) and Working Groups
- Establish AI Community of Practice
- Develop NRC-specific language model for Natural Language Processing (NLP)
- Optimize NRC software intake and approval process for AI tools
- Coordinate with NRC Data Strategy to target cloud development and deployment environment
- Continue Data Science and AI Regulatory Applications Workshops
- Host internal RES AI Seminars (next planned August 2022)
RECENT NRC AI RESEARCH PROJECTS Dr. Trey Hathaway Reactor Systems Engineer (Code Development)
Office of Nuclear Regulatory Research
USING AI TO ENHANCE NRC ACTIVITIES Administrative Conference of the United States Statement on Agency Use of Artificial Intelligence Evidence-Based Policymaking Act of 2018 National Artificial Intelligence Initiative Act of 2020 Office of Management and Budget Guidance for Regulation of AI Applications (M-21-06)
AI Strategic Plan outputs may also support the agency use of AI tools to enhance internal NRC activities
Resource Estimation
- Challenge: Deviations between resource estimates to complete a licensing review and the actual hours charged
- Goal: Create tool to assist project managers in formulating resource estimates
- Leverage historical data
- Find historically similar reviews
- Method: Use term frequency-inverse document frequency vectors to represent documents and perform similarity calculations
- Rank documents based on similarity
Term Frequency-Inverse Document Frequency (Vector Representation)
Represent a document as a vector
- The vector reflects the word usage in the document
- The vector will have 1000s of dimensions wordx wordz action amendment applicability assembly base Document 1 0.0 0.16 0.2 0.00 0.33 storage system technical time wording would 0.0 0.15 0.3 0.22 0.00 0.25
Term Frequency-Inverse Document Frequency (Vector Space Corpus)
Represent the collection of documents as vectors Create a vocabulary of all words used in the collection wordx wordz action amendment applicability assembly base Document 1 0.0 0.16 0.2 0.00 0.33 Document 2 0.25 0.0 0.0 0.3 0.0 storage system technical time wording would 0.0 0.15 0.3 0.22 0.00 0.25 0.11 0.0 0.0 0.3 0.14 0.1
Term Frequency-Inverse Document Frequency (Similarity Calculations)
A new document is converted to a vector based on the vocabulary of the collection of documents
- The similarity (angle between vectors) is calculated as the dot product between vectors
- Documents ranked by similarity score wordx wordz
Resource Estimation Tool
Resource Estimation Tool
Current Status and Follow-on Work
- Preliminary acceptance testing complete
- Historical data provides reasonable estimates of required resources and review durations
- Coordinating to finalize visualizations with NRC program offices
- Develop and deploy final User Interface
- Potential Follow-on Work:
- Search capabilities
- Predict Branch assignments
- Predict Standard Review Plan
- Predict which Regulatory Guides were used for the licensing action
Regulatory Named Entity Recognition
- Challenge: Title 10 of the Code of Federal Regulations (CFR), and other regulatory documents, reference sections of 10 CFR
- Revisions to 10 CFR could impact other sections
- Goal: Create a tool to find and extract 10 CFR references from documents
- Method: Use Named Entity Recognition (NER) to label text as regulations and extract that text
- Named Entity Recognition
-Classify unstructured text (i.e.,
words or phrases) into predefined categories
- Information extraction
- Typical tools (e.g. spaCy) trained on general language
SpaCy Default Entities Addition of NRC Specific Language Patterns Named Entity Recognition Used Python package Spacy
10 CFR Reference Identification Tool
10 CFR Reference Identification Tool
10 CFR Reference Identification Tool
Contact Us Matt Dennis Reactor Systems Engineer (Data Scientist)
Office of Nuclear Regulatory Research matthew.dennis@nrc.gov Dr. Trey Hathaway Reactor Systems Engineer (Code Development)
Office of Nuclear Regulatory Research alfred.hathaway@nrc.gov Reed Anzalone (NRR) - NRR Liaison Andrew Lerch (NRR/EVS) - NRR Liaison