State of Charge and State of Health Estimation for Lithium Batteries Using Recurrent Neural Networks

• Published in: IEEE transactions on vehicular technology Vol. 66; no. 10; pp. 8773 - 8783
• Main Authors: Chaoui, Hicham, Ibe-Ekeocha, Chinemerem Christopher
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.10.2017; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Aging; Ambient temperature; Batteries; Discharges (electric); Electric vehicles; Estimation; Intelligence; lifetime estimation; Lithium; Lithium batteries; neural networks; Power management; Rechargeable batteries; Recurrent neural networks; Robustness (mathematics); State of charge; state of health; Temperature measurement; Vehicle dynamics; Voltage measurement
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2017.2715333

This paper presents an application of dynamically driven recurrent networks (DDRNs) in online electric vehicle (EV) battery analysis. In this paper, a nonlinear autoregressive with exogenous inputs (NARX) architecture of the DDRN is designed for both state of charge (SOC) and state of health (SOH) estimation. Unlike other techniques, this estimation strategy is subject to the global feedback theorem (GFT) which increases both computational intelligence and robustness while maintaining reasonable simplicity. The proposed technique requires no model or knowledge of battery's internal parameters, but rather uses the battery's voltage, charge/discharge currents, and ambient temperature variations to accurately estimate battery's SOC and SOH simultaneously. The presented method is evaluated experimentally using two different batteries namely lithium iron phosphate (LiFePO 4 ) and lithium titanate (LTO) both subject to dynamic charge and discharge current profiles and change in ambient temperature. Results highlight the robustness of this method to battery's nonlinear dynamic nature, hysteresis, aging, dynamic current profile, and parametric uncertainties. The simplicity and robustness of this method make it suitable and effective for EVs' battery management system (BMS).