TY - JOUR
T1 - Developing a real-time data-driven battery health diagnosis method, using time and frequency domain condition indicators
AU - Khaleghi, Sahar
AU - Firouz, Yousef
AU - Van Mierlo, Joeri
AU - Van Den Bossche, Peter
PY - 2019/12/1
Y1 - 2019/12/1
N2 - Lithium-ion batteries are considered as promising electric energy storage systems. However, identification of battery health is a critical issue. Furthermore, battery aging extremely depends on operating conditions. Therefore, monitoring and analysis of battery health degradation in real-time systems such as electric vehicles, in which a variety of stress factors may come into play, are demanded. This paper proposes a data-driven algorithm based on multiple condition indicator to estimate battery health using application-based load profiles. In this regard, battery cells have been cycled under a worldwide light duty driving test cycle (WLTC) load profile in laboratory to acquire real-world driving data. Time-domain and frequency-domain condition indicators are extracted from measured on-board data like voltage and current within certain time intervals, enabling real-time investigation of battery health degradation. The condition indicators have been fed into a Gaussian process estimator to track the real-time state of health (SoH). As degradation strongly depends on magnitude of input current, it is important that the proposed method can predict health of the cell regardless of current amplitude and aging pattern. Therefore, to assess accuracy and robustness of the proposed method, it is validated using a different load profile with distinct depth of discharge, current amplitude, and distinctive aging pattern. Results reveal the proposed approach is highly precise and is capable of estimating battery SoH with low computational costs and a relative error of less than 1%. The proposed technique is promising for online diagnostics of battery health thanks to its high accuracy and robustness.
AB - Lithium-ion batteries are considered as promising electric energy storage systems. However, identification of battery health is a critical issue. Furthermore, battery aging extremely depends on operating conditions. Therefore, monitoring and analysis of battery health degradation in real-time systems such as electric vehicles, in which a variety of stress factors may come into play, are demanded. This paper proposes a data-driven algorithm based on multiple condition indicator to estimate battery health using application-based load profiles. In this regard, battery cells have been cycled under a worldwide light duty driving test cycle (WLTC) load profile in laboratory to acquire real-world driving data. Time-domain and frequency-domain condition indicators are extracted from measured on-board data like voltage and current within certain time intervals, enabling real-time investigation of battery health degradation. The condition indicators have been fed into a Gaussian process estimator to track the real-time state of health (SoH). As degradation strongly depends on magnitude of input current, it is important that the proposed method can predict health of the cell regardless of current amplitude and aging pattern. Therefore, to assess accuracy and robustness of the proposed method, it is validated using a different load profile with distinct depth of discharge, current amplitude, and distinctive aging pattern. Results reveal the proposed approach is highly precise and is capable of estimating battery SoH with low computational costs and a relative error of less than 1%. The proposed technique is promising for online diagnostics of battery health thanks to its high accuracy and robustness.
UR - http://www.scopus.com/inward/record.url?scp=85072026175&partnerID=8YFLogxK
U2 - 10.1016/j.apenergy.2019.113813
DO - 10.1016/j.apenergy.2019.113813
M3 - Article
VL - 255
JO - Applied Energy
JF - Applied Energy
SN - 0306-2619
M1 - 113813
ER -