Machine-Learning-Based Efficient Parameter Space Exploration for Energy Storage Systems

Tuesday, November 26, 2024
11:00 a.m.-12:00 p.m.
Kay Boardroom - Kim Engineering Building
Catherine Stephens
301 405 9378
csteph5@umd.edu

Machine-Learning-Based Efficient Parameter Space Exploration for Energy Storage Systems

There is considerable interest in developing new energy storage technologies for the electric grid, but economic viability will require that manufacturers provide warranties guaranteeing 15-20+ years of life. Although there are extensive efforts to make early predictions for the expected life of new storage technologies, we argue that knowing the expected life has little or no commercial value. Instead, full failure probability distribution is required. Here, we develop a framework based on Gaussian processes, equipped with domain knowledge, to implement a Bayesian optimization approach to explore the parameter space efficiently and quantify durability using failure probability distributions. Our experimental results show accurate durability predictions with a significantly reduced number of experiments. We also propose how to use our approach to accurately predict durability under conditions where the charging and discharging profiles are complex functions of time.

--------------------------

Steve Harris received a BS degree in chemistry from UCLA and a PhD in physical chemistry from Harvard University. After a Miller Post-Doctoral Fellowship at UC Berkeley, he began his career at the General Motors Research Labs. Apart from a 9 year stint at the Ford Scientific Research Labs, Steve worked at GM as a Technical Fellow until 2011, when he was awarded a Miller Visiting Professorship in the UC Berkeley Chemistry Department. Since then, he has worked in the Materials Science and Energy Storage Divisions at Lawrence Berkeley Lab. During that time, he has been a Visiting Scholar in the Materials Science and Engineering Department at Stanford, and he has consulted for battery companies, private equity companies, and venture capital companies. His most recent interest involves using machine learning to predict and understand battery durability. His h-index is 72, and he has more than 20,000 citations. He is also on the advisory boards of several battery startups. Steve has mentored over 50 young scientists in the past year, talking about career choices, brainstorming, green cards, and anything else. 

 

Audience: Graduate  Faculty 

remind we with google calendar

 

December 2024

SU MO TU WE TH FR SA
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30 31 1 2 3 4
Submit an Event