An execution time optimized state of charge estimation method for lithium-ion battery

State of charge (SOC) estimation is one of the most important outputs of the battery management system (BMS). There are several algorithms for SOC estimation. Each method has advantages like self-correction ability and disadvantages like computational complexity for an embedded system. In the first part of this article, different estimation techniques are reviewed and compared. Particular emphasis was placed on two methods: Coulomb Counting (CC) and Sigma Point Kalman Filter (SPKF). These two methods are analyzed in terms of several aspects such as tolerance to noisy signal, recovery ability from an intentional SOC distortion as well as estimation accuracy comparison. Also, an embedded development kit is used to analyze execution time of each method. The results show that using SPKF alone is a computationally expensive method especially for a battery pack with a high number of cells in series. On the other hand, using CC alone could be vulnerable to SOC distortion and noisy measurement signals and this may lead to inaccurate predictions. Based on these facts, a hybrid approach has been proposed to take advantage of each method's superiority. The emphasis in this paper is on execution step time analyses under different conditions and drive cycles. On top of this article, further studies may pave the way for more cost effective solutions like downsizing of the processor.

Automated summary

This article reviews and compares different algorithms for state of charge (SOC) estimation in the battery management system (BMS). It focuses on Coulomb Counting (CC) and Sigma Point Kalman Filter (SPKF) methods and proposes a hybrid approach to leverage their advantages. The results show that using SPKF alone is computationally expensive, while using CC alone may lead to inaccurate predictions. The article also analyzes execution time and suggests the possibility of downsizing the processor.