Search

Accurate state of charge (SOC) estimation provides an essential basis for the functionalities of battery management systems in electric vehicles (EVs). However, conventional equivalent circuit models suffer significant model accuracy deterioration under extreme SOCs, nonroom temperatures, and heavy loads. In this work, we implant the Butler–Volmer (BV) equation and the fractional-order model representation into a model-based physics-informed neural network (M-PINN) to simulate current-dependent battery charge transfer dynamics under various operating conditions. This M-PINN replaces the original neuron structure with a set of submodels and allows the BV coefficient to be randomly selected in a roughly estimated range for each submodel. By applying the Lyapunov analysis, a self-adaptive neural network-based fractional-order observer is proposed to guarantee the uniform ultimate boundedness stability of both system states and M-PINN weights, thereby achieving accurate online SOC estimation without necessitating substantial data and efforts for offline neural network training. The experimental validations are implemented under three EV driving profiles with different average currents at −5, 5, 20, and 35 Celsius. The validation results demonstrate that the proposed method achieves the mean absolute errors of less than 0.9% in all the validation scenarios.

This article proposes an information analysis-based multiple adaptive forgetting factors (FFs) recursive least squares (IA-MAFF-RLS) method to identify model parameters of lithium-ion batteries in electric vehicles. First, a Cramer–Rao lower bound (CRLB) based information analysis is implemented for each individual model parameter, and the constant information theory is introduced to update the CRLB for the associated estimation error covariance. Second, a switching-based adaptive strategy is proposed with two rules to fine-tune multiple FFs online by making a trade-off between the information richness in a memory window and the traceability to current-voltage profiles. Third, the IA-MAFF-RLS method is established to separately identify different model parameters with associated FFs. Unlike conventional methods, the proposed element-wise forgetting strategy directly focuses on battery model parameters rather than regression model coefficients, thereby avoiding an accuracy loss in the transformation between two models. According to the simulation and experimental results, the proposed method reduces the mean square deviation to −20.17 dB under noise interferences and benefits the generic extended Kalman filter in online SOC estimation with the estimation error less than 1% at different battery aging levels and temperatures.

State of power (SOP) reflects the peak power capability of a lithium-ion battery (LIB). Constant power (CP) operation (e.g., discharge or charge) is more representative of actual battery loadings in electric vehicle (EV) applications (e.g., EV acceleration, gradient climbing and regenerative braking) than constant current or constant voltage operation. However, relevant research on CP-based SOP estimation for LIBs in EVs is still rare. In this article, a novel model switching-based iterative algorithm is proposed for multi-constraint SOP estimation under a CP operation scenario. Two state-space models with implicit representations are constructed to describe the indirect relationships between a given CP over a prediction window and the maximum look-ahead current/voltage at the end of this window. An unscented Kalman filter-based correction strategy is applied to determine the dominant limitation factor so as to pinpoint battery SOP in high fidelity. Moreover, a SOP testing approach is designed to calibrate reference SOPs in high precision for validations. Experimental results have demonstrated that the proposed method achieves promising accuracy with a mean absolute error of less than 0.26 W under both static and dynamic conditions.

Soft short circuit (SC) fault diagnosis is critical for a battery management system to prevent thermal runaway of lithium-ion batteries in electric vehicles. In this article, an adaptive neural network based observer is developed to estimate soft SC faults based on an augmented battery model subject to unknown nonlinear uncertainties. Rigorous theoretical analysis in terms of estimation error convergence is then provided. Leveraging the designed observer, a delicate diagnosis algorithm is presented to timely detect SC fault occurrences and an iterative updating method is further applied to accurately estimate the SC fault resistance via thresholding. Finally, comprehensive experimental tests and comparative studies are elaborated to validate the effectiveness and superiority of the proposed algorithm.

This article proposes a curve relocation approach for robust battery open circuit voltage (OCV) reconstruction and capacity estimation based on partial charging data. First, an electrode-level aging mechanism analysis is conducted to reveal the underlying reasons for battery OCV distortion and capacity decay, and three electrode aging parameters (EAPs) are proposed to account for those aging-induced relative position shifts of electrode OCV curves. Second, a deep long short-term memory recurrent neural network with a many-to-one structure is established to yield battery OCV estimations in high fidelity using accessible daily charging data. Then, a multicoupled optimization algorithm is designed to accurately estimate EAPs, which ensures that the reconstructed OCV curves match well with the estimated OCV segments while satisfying various OCV-related health features at a specific aging level. Obtaining the optimal EAPs contributes to 1) relocate the relative positions of electrode OCV curves for reliable battery OCV reconstruction; 2) determine battery usable capacity range and achieve capacity estimation in high accuracy; and 3) help understand electrode aging behaviors. The proposed method realizes a mean absolute error of less than 20 mV in battery OCV reconstruction and a mean absolute percentage error of less than 1.3% in capacity estimation given a short charging segment up to 1000 s over the whole battery lifetime.