Successive-approximation algorithm for estimating capacity of Li-ion batteries. (15th September 2018)

• Record Type: Journal Article
• Authors: Goh, Taedong; Park, Minjun; Seo, Minhwan; Kim, Jun Gu; Kim, Sang Woo
• Is Part Of: Energy. Volume 159(2018)
• Page Start: 61
• Page End: 73
• Publication Date: 2018-09-15
• Subjects: State of health (SOH) -- Fixed point iteration -- Cycle aging -- Capacity estimation -- Lithium; 00-01 -- 99-00; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2018.06.116
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: This paper proposes a capacity estimation algorithm for Li-ion batteries (LIBs) using the successive approximation method. Model-based capacity estimation method can be applied to a variety of current profiles because the capacity is calculated from state of charge (SOC) estimated accurately using the functional relationship between open circuit voltage (OCV) and the SOC. However, with aging, the OCV-SOC table changes, which worsens the estimation accuracy. Therefore, additional experiments are necessary to compensate the errors whenever the capacity should be identified. To overcome this restriction, this paper proposes an algorithm for estimating both the capacity and the corresponding OCV-SOC table based on the preliminary OCV-SOC tables obtained from other batteries. The capacity and the table are updated successively based on the prior capacity estimate. This work proposes two algorithms for voltage characteristics: OCV measurement and SOC estimation cases. The former uses the measured OCV to calculate the SOCs directly, while the latter estimates the SOCs using a dual extended Kalman filter (DEKF). Aging data from five LIB packs are analyzed, and the capacity estimation errors are less than 2.2% for the OCV measurement case and 3.06% until 20% loss of capacity estimate for SOC estimation case. Highlights: A new algorithm estimates capacity and OCV-SOC table iteratively using aging data. On-board tests for updating the OCV-SOC table are not required. Capacity can be estimated for both of OCV measurement and SOC estimation cases. Data of differently aged batteries were analyzed to verify the proposed algorithm. Capacity estimation errors were less than 3.06% up to 20% loss of capacity estimate.