2019 Theses Master's
Battery State-of-Charge Estimation Using Neural Networks
This thesis proposes a way to augment the existing machine learning algorithm applied to state-of-charge estimation by introducing a form of pulse injection to the running battery cells. It is believed that the information contained in the pulse responses can be interpreted by a machine learning algorithm whereas other techniques are difficult to decode due to the nonlinearity. OCV Mapping is also applied in order to evaluate and compare the performances with feedforward neural network (FNN) -based approach. Coulomb-counting is selected as the basis of making comparison as it is capable of obtaining the SoC with an error less than 0.1%. The detailed system layout is given to perform the augmented SoC estimation integrated in a real-world testbench. Testing procedures specifically designed for both OCV Mapping and FNN-based approach are also explained and provided. A 2-hidden layer FNN is trained to acquire the nonlinear relationship between the training pulse and the ground-truth SoC. The experimental data is trained and the results are shown within 6-8mins computation time and an error boundary of 1.13% for charge and 0.80% for discharge, whereas OCV Mapping has approximately 3.35% SoC estimation error for charge and 1.86% for discharge even after 90 minutes relaxation.
Subjects
Files
- Chenyao_Liao_masters thesis_cl3757.pdf application/pdf 2.29 MB Download File
More About This Work
- Academic Units
- Electrical Engineering
- Thesis Advisors
- Preindl, Matthias
- Degree
- M.S., Columbia University
- Published Here
- May 12, 2020
Notes
Key words: Battery Balancing, Battery Management Systems (BMSs), Li-ion batteries, SoC Estimation, Machine Learning, Neural Network.