Online state-of-health estimation of lithium-ion batteries using Dynamic Bayesian Networks

• Published in: Journal of power sources 2014-12, Vol.267, p.576-583
• Main Authors: He, Zhiwei, Gao, Mingyu, Ma, Guojin, Liu, Yuanyuan, Chen, Sanxin
• Format: Article
• Language: English
• Subjects: Applied sciences; Battery management system; Bayesian analysis; Direct energy conversion and energy accumulation; Dynamic Bayesian Network; Dynamical systems; Dynamics; Electric batteries; Electrical engineering. Electrical power engineering; Electrical power engineering; Electrochemical conversion: primary and secondary batteries, fuel cells; Exact sciences and technology; Lithium batteries; Lithium-ion batteries; Lithium-ion battery; Networks; Online; State of health; Storage batteries
• ISSN: 0378-7753; 1873-2755
• DOI: 10.1016/j.jpowsour.2014.05.100

Li-ion batteries are widely used in energy storage systems, electric vehicles, communication systems, etc. The State of Health (SOH) of batteries is of great importance to the safety of these systems. This paper presents a novel online method for the estimation of the SOH of Lithium (Li)-ion batteries based on Dynamic Bayesian Networks (DBNs). The structure of the DBN model is built according to the experience of experts, with the state of charges used as hidden states and the terminal voltages used as observations in the DBN. Parameters of the DBN model are learned based on training data collected through Li-ion battery aging experiments. A forward algorithm is applied for the inference of the DBN model in order to estimate the SOH in real-time. Experimental results show that the proposed method is effective and efficient in estimating the SOH of Li-ion batteries. •A novel SOH estimation method based on Dynamic Bayesian Networks is proposed.•The SOH can be estimated in an online manner.•Only terminal voltages during the constant charge process should be measured.•The estimated SOH can be provided inherently as either a fuzzy or an exact value.