Online modelling and state-of-charge estimation for lithium-titanate battery

• Main Author: Low, Wen Yao
• Format: Thesis
• Language: English
• Published: 2016
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utm.my/id/eprint/84073/1/LowWenYaoPFKE2016.pdf; http://eprints.utm.my/id/eprint/84073/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:126334
• Institution: Universiti Teknologi Malaysia

Superior safety, is a promising energy storage element for electric vehicles. Its features can be fully utilised by using a fast charger and a high performance battery management system. Battery model is vital to a battery charger design for characterising the charging behaviours of a battery. Additionally, a robust state-ofcharge (SoC) estimation should be realised for a reliable battery management. This thesis develops a battery model for charger design and a robust method for SoC estimation by using MATLAB. The thesis proposed a transfer function-based battery model which is applicable for small-signal analysis and large-signal simulation of battery charger design, in order to capture the charging profiles of LTO battery. Busse’s adaptive rule, which has simple computations, is applied to improve the accuracy of Kalman filter-based SoC estimation. Busse’s adaptive Kalman filters are also applied for SoC estimation with online battery modelling to eliminate the complicated process of battery modelling. This study was conducted by using 2.4 V, 15 Ah LTO batteries. The batteries were tested with continuous current test and pulsed current test at several ambient temperatures (-25 ºC, 0 ºC, 25 ºC and 50 ºC) and charge/discharge currents (0.5 C, 1 C, 2 C). Additionally, modified dynamic stress tests at several temperatures (-15 ºC, 0 ºC, 15 ºC, 25 ºC, 35 ºC and 50 ºC) were also performed to test the battery under real EV environment. Results of the battery modelling showed that the developed transfer function-based battery model is accurate where the root-mean-square modelling error is less than 30 mV. The results also revealed that the Busse’s adaptive rule has effectively improved the Kalman filter-based SoC estimation for the case of offline battery model by giving a higher accuracy and shorter convergence time. Additionally, Busse’s adaptive Extended Kalman Filter gave a better accuracy in SoC estimation with online battery modelling. The proposed transfer function-based battery model provides a helpful solution for the battery charger design while the proposed Busse’s adaptive Kalman filter offers an accurate and robust SoC estimation for both offline and online battery models.