Prediction of Battery Health Dynamics

<p>"The successful market penetration of lithium-ion batteries in the last 30 years has ushered in a new dawn in energy applications. Notwithstanding their widespread use, lithium-ion batteries are affected by a plethora of fundamental issues — safety, cycle life, performance, and cost. In this work, advanced modeling techniques are developed to tackle some of these issues. This dissertation is subdivided into four parts, the first is introduction while the remaining are dedicated to the development of computational techniques for predicting battery health dynamics.</p> <p>In the second part of this work, we developed fast and accurate algorithms for modeling essential battery states. These algorithms employed the Chebyshev spectral method, which guaranteed 51% - 98% reduction in computation time and accuracy that was well within 0.07% - 0.39% of a high-fidelity COMSOL reference. In the third part of this work, we extended the earlier developed Chebyshev spectral algorithm to capture degradation physics in a lithium-ion battery. The modeled degradation physics are — solid electrolyte film formation, hydrogen gas evolution, Mn deposition, salt decomposition, Mn dissolution, solvent oxidation, and reduction in diffusion coefficient. Our implementation led to 91% reduction in time, and accuracy that was well with 0.1358% - 0.28% of the reference model. In the last part of this work, a machine learning model is developed for monitoring the state-of-health of a battery. For the generation of training and test dataset, we relied on a physics-based degradation model. The resulting accuracy of our state-of-health prediction was within 0.004249% of the reference physics-based model, while its computation time takes about 3s, as opposed to the reference model which takes ~3hrs"--Abstract, p. iv</p>

Name thesis:degree_name

Ph. D. in Mechanical Engineering

Grantor

Missouri University of Science and Technology

Author dc:creator

• Ajiboye, Damola Martins

• Chebyshev spectral modeling
• Data driven battery modeling
• Energy storage for microgrids
• Fast and accurate battery modeling
• Online estimation
• Physics-based battery modeling
• Engineering
• Mechanical Engineering

Repository record dc:identifier

https://scholarsmine.mst.edu/doctoral_dissertations/3187

OAI identifier oai:identifier

oai:scholarsmine.mst.edu:doctoral_dissertations-4192