Open Access

ARTICLE

Lithium-Ion Battery Screening by K-Means with DBSCAN for Denoising

Yudong Wang^{1, 2}, Jie Tan^{1, *}, Zhenjie Liu¹, Allah Ditta³

1 Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China.
2 School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, 100049, China.
3 University of Education, Township, College Rd Lahore Punjab, Lahore, 54770, Pakistan.

* Corresponding Author: Jie Tan. Email: .

Computers, Materials & Continua 2020, 65(3), 2111-2122. https://doi.org/10.32604/cmc.2020.011098

Received 19 April 2020; Accepted 20 June 2020; Issue published 16 September 2020

Abstract

Batteries are often packed together to meet voltage and capability needs. However, due to variations in raw materials, different ages of equipment, and manual operation, there is inconsistency between batteries, which leads to reduced available capacity, variability of resistance, and premature failure. Therefore, it is crucial to pack similar batteries together. The conventional approach to screening batteries is based on their capacity, voltage and internal resistance, which disregards how batteries perform during manufacturing. In the battery discharge process, real time discharge voltage curves (DVCs) are collected as a set of unlabeled time series, which reflect how the battery voltage changes. However, few studies have focused on DVC based battery screening. In this paper, we provide an effective approach for battery screening. First, we apply interpolation on DVCs and give a method to transform them into slope sequences. Then, we use density-based spatial clustering of applications with noise (DBSCAN) for denoising and treat the remaining data as input to the K-means algorithm for screening. Finally, we provide the experimental results and give our evaluation. It is proved that our method is effective.

---

Keywords

---