Project Info
Generative AI for enhancing energy system planning and enabling energy transformation
Qiuhua Huang
qiuhuahuang@mines.edu
Project Goals and Description:
Renewable energy has a great potential to meet the increasing energy demand while addressing the climate change crisis. There are over 2000 GW renewable energy projects (more than all existing generation resources in U.S.) proposed and waiting in the interconnection queque for years, due to lack of resources and inefficiency in the energy planning process. This project will explore and develop an <b>intelligent co-planner</b>, named LLM-planner, based on Large Language Model (LLM) technologies and the state-of-the-art open-sourced LLM models to <u>assist human transmission planners to accomplish comprehensive planning and renewable interconnection studies much more efficiently</u>.
<b>Motivation</b>
<ul>
<li>Electric power system planning is essential for ensuring system reliability while supporting large-scale renewable integration. Existing planning processing is very labor-intensive and too slow to meet the industry need.</li>
<li>Many planning tasks heavily relies on planner’s experience and knowledge accumulation. Transfer of these insights and intuitions poses a formidable challenge after experienced planners retire.</li>
<li>A comprehensive Artificial Intelligent based solution, akin to an intelligent co-planner, capable of delivering task automation and informed recommendations in planning process, remains absent.</li>
</ul>
<b>Project Approach and Expected Outcomes </b>
<ul>
<li>Knowledge retrieval module that provides the LLM-Planner specific contexts about planning procedure and practices through retrieving planning knowledge database and documents.</li>
<li>Tool and code execution module that takes commands from the reasoning and planning module and executes appropriate planning tools or other plugin AI agents through their APIs to complete subtasks.</li>
</ul>
More Information:
Grand Challenge: Make solar energy economical.
News reports about the interconnection and transmission planning problems:
<ul>
<li>https://www.nytimes.com/interactive/2023/06/12/climate/us-electric-grid-energy-transition.html</li>
<li>https://energynews.us/2023/05/30/a-line-too-long-grid-interconnection-delays-threaten-states-clean-energy-goals/</li>
</ul>
U.S. Department of Energy interconnection roadmap:
<ul>
<li>https://www.energy.gov/eere/i2x/doe-transmission-interconnection-roadmap-transforming-bulk-transmission-interconnection</li>
</ul>
Generative AI and its potential for energy transformation
<ul>
<li>https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai</li>
<li>https://www.utilitydive.com/news/utilities-generative-ai-artificial-intelligence-capgemini-report/686601/</li>
</ul>
Primary Contacts:
Qiuhua Huang, qiuhuahuang@mines.edu
Student Preparation
Qualifications
The most important qualification is motivation and passion to solve real-world energy/climate change problems. In addition, the student is expected to have a good background in programming or computer science and AI, plus some general background in electrical engineering/power systems or physics. Thus an ideal candidate will be EE students with a CS/CE minor. CS students with passion to solve energy and climate problems are also welcome. Prior experience with generative AI is a big plus.
TIME COMMITMENT (HRS/WK)
4
SKILLS/TECHNIQUES GAINED
<ul>
<li>Problem solving skills: identifying the problem and potential solutions through reading papers, reports and learning from your advisors, implement solutions, test it, reflect and iterate to move forward</li>
<li>Learning the state of the art generative AI methods and tools</li>
<li>Hands-on experience on the state-of-the art open-sourced LLM models such as Llama 3 for power and energy system planning and interconnection study applications</li>
<li>Oral and written communication skills, for example, the project outcomes will be summarized as a conference paper with the student as the first author.</li>
</ul>
MENTORING PLAN
The student will meet with the advisor and phD students on a bi-weekly schedule; Additional meetings could be scheduled as needed to help address issues/roadblocks in the project.
Preferred Student Status
Junior
Senior