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

Aug 2, 2026

Upload by Levi

Energy storage systems must estimate remaining energy across their lifetime, yet this depends on many interacting operating parameters, creating a large, high-dimensional parameter space. Exhaustive testing on dense grids is prohibitively costly, further amplified by cell-to-cell performance variability. This work develops a Gaussian-process framework using domain knowledge to enable Bayesian optimization, selecting experiments that maximize information gain while reducing uncertainty. Results demonstrate accurate remaining-energy prediction with far fewer experiments. To address laboratory-to-field mismatch, an approach is proposed to predict performance under real-world cycling conditions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages28

Reading Time1 h 11 min

AccessFree to download

Abstract

Introduction

Remaining Energy Estimation and Parameter Space Challenge

Parameter-Dependent Degradation Modeling

Physics-Based and Machine-Learning Approaches

Lawrence Berkeley National Laboratory LBL Publications Title Machine-Learning-Based Efficient Parameterspace Exploration for Energy Storage Systems Permalink [https://escholarship.org/uc/item/62b4h7vb](https://escholarship.org/uc/item/62b4h7vb) Journal ECS Meeting Abstracts, MA2024-02(3) ISSN 2151-2043 Authors Harris, Stephen J Alghalayini, Maher Noack, Marcus et al. Publication Date 2024-11-22 DOI 10.1149/ma2024-023394mtgabs Peer reviewed [eScholarship.org](eScholarship.org) Powered by the California Digital Library University of California Machine-Learning-Based Efficient Parameter Space Exploration for Energy Storage Systems Maher B. Alghalayini 1,2*, Daniel Collins-Wildman 1 , Kenneth Higa 1 , Armina Guevara 1 , Vincent Battaglia 1 , Marcus M. Noack2*, Stephen J. Harris 1 1 Energy Storage and Distributed Resources Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, 94720, California, USA. 2 Applied Mathematics and Computational Research Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, 94720, California, USA. *Corresponding author(s). E-mail(s): [MAlghalayini@lbl.gov](MAlghalayini@lbl.gov) ; [MarcusNoack@lbl.gov](MarcusNoack@lbl.gov) ; Contributing authors: [dcollinswildman@lbl.gov](dcollinswildman@lbl.gov) ; [khiga@lbl.gov](khiga@lbl.gov) ; [arminamguevara@berkeley.edu](arminamguevara@berkeley.edu) ; [vsbattaglia@lbl.gov](vsbattaglia@lbl.gov) ; [SJHarris@lbl.gov](SJHarris@lbl.gov) ; Abstract Shifting towards sustainable energy sources requires developing new storage systems and estimating their remaining energy over their lifetime. The remaining energy of these systems depends on many operating parameters, resulting ina large high-dimensional parameter space to explore. Testing cells exhaustivelyon a dense grid in the parameter space is prohibitively expensive. This is especially true with considerable cell-to-cell variability in performance, even under the same cycling conditions. Here, we develop a framework based on Gaussian processes, equipped with domain knowledge, to implement Bayesian optimization to explore the parameter space efficiently and quantify remaining energy using failure distributions. Bayesian optimization identifies future experiments that maximize information gain and minimize uncertainty. Experimental results show accurate remaining energy predictions with significantly fewer experiments. However, laboratory cycling conditions, including those in literature, may not represent real-world cycling. We propose an approach based on laboratory results to predict remaining energy under real-world cycling conditions. 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050 051 052 053 054 055 056 057 058 059 060 061 062 063 064 065 066 067 068 069 070 071 072 073 074 075 076 077 078 079 080 081 082 083 084 085 086 087 088 089 090 091 092 093 094 095 096 097 098 099 100 Keywords: Machine Learning, Bayesian Optimization, Gaussian Process, Stochastic Modeling, Parameter Space, Efficient Exploration, Energy Storage, Battery Failure Modeling, Complex Cycling Conditions 1 INTRODUCTION In the fight against global warming, the demand for new energy storage technologies has increased dramatically. Although solar and wind have great potential to fight global warming, these intermittent sources hinder grid integration. Energy storage systems have risen as a popular solution [1] . Within the grid, energy storage systems store excess energy during peak generation periods and release energy when needed during low-energy generation periods. While commercially availa

FAQs

Why is exploring the energy storage parameter space difficult?

Remaining energy depends on many operating and cell parameters, and all combinations form a very high-dimensional space. Exhaustive grid testing is prohibitively expensive, especially under cell-to-cell variability.

What method does the document use to explore parameters efficiently?

It uses a Gaussian-process framework combined with Bayesian optimization, enhanced with domain knowledge. The approach selects future experiments to maximize information gain and minimize uncertainty.

How is remaining energy quantified in the proposed framework?

The framework quantifies remaining energy using failure distributions. Bayesian optimization is then used to efficiently explore parameters that influence those distributions.

Aug 2, 2026

Upload by Levi