(584h) Model-Based Stochastic Fault Detection and Diagnosis of Lithium-Ion Batteries

Lithium-ion battery (Li-ion) is becoming the dominant energy storage solution in hybrid electric and electric vehicles, due to its higher energy density and longer life cycle [1]. For these applications, the battery must perform reliably and pose no safety threats. However, the performance of the Li-ion battery can be affected by abnormal thermal behaviors, which are defined as faults. For example, as previously reported, thermal runaway can cause fire, and temperature imbalances in battery cells can affect the aging behavior of battery [2, 3]. It is essential to develop reliable thermal management system to accurately predict and monitor thermal behaviors of Li-ion battery [4]. Using the first-principle models of Li-ion battery, this work presents a stochastic fault detection and diagnosis (FDD) algorithm to identify two particular faults in the Li-ion battery cell, i.e., thermal runaway fault and convection fault.
Models of Li-ion battery are typically derived from the underlying physical phenomena. However, to make model useful and tractable, it is common to make assumptions and simplification during the model development, which consequently may introduce mismatch between the model and the battery under study. An iterative real-time optimization is developed in this work, which can correct the model by incorporating the error between measurements and model prediction of temperatures.
The performance of the FDD algorithm can be affected by uncertainty, which may originate from intrinsic time varying phenomena or may results from model calibration with noisy data [5]. To evaluate the effect of uncertainty on FDD, a generalized polynomial chaos (gPC) expansion is used to firstly approximate uncertainty in thermal dynamics and then propagate such an uncertainty onto model predictions that can be used for FDD. As compared to sampling-based uncertainty analysis techniques such as the Monte Carlo (MC) simulations, the gPC-based approximation of uncertainty in thermal dynamics can significantly reduce the computational complexity and improve the accuracy of FDD [6].
Additionally, most of the available FDD algorithms in the literature use fault signatures identified from a pattern recognition method or residual calculated from an observer to detect the occurrence of faults [7, 8]. These techniques provide little information about the probability that a fault has occurred. Since faults occurring in the battery may be of stochastic nature and since the uncertainty in thermal dynamic behaviors is considered in this work, the use of fault signature or residual for FDD may not be effective. A joint confidence region (JCRs)-based optimization is developed for stochastic fault detection, which can provide a probabilistic description of the occurrence of faults.
The model correction and stochastic fault detection algorithms are applied to a two-state Li-ion battery system with stochastic thermal faults that can affect the core and surface temperatures of battery. The efficiency of the proposed FDD algorithm is quantified in terms of fault detection rate for both individual and simultaneous faults in Li-ion battery.

References

[1] J. M. Tarascon and M. Armand, "Issues and challenges facing rechargeable lithium battries," Nature, vol. 414, no. 6861, pp. 359-367, 2001.

[2] S. Dey, Z. A. Biron, S. Tatipamula, N. Das, S. Mohon, B. Ayalew and P. Pisu, "Model-based real-time thermal fault diagnosis of Lithium-ion batteries," Control Engineering Practice, vol. 56, pp. 37-48, 2016.

[3] Q. Wang, P. Ping, X. Zhao, G. Chu, J. Sun and C. Chen, "Thermal runaway caused fire and explosion of lithium ion battery," Journal of Power Sources, vol. 208, pp. 210-224, 2012.

[4] M. Charkhgard and M. Farrokhi, "State-of-Charge Estimation for Lithium-Ion Batteries Using Neural Networks and EKF," IEEE Transactions on Industrial Electronics, vol. 57, no. 12, pp. 4178-4187, 2010.

[5] Y. Du, H. Budman and T. Duever, "Fault detection and diagnosis with parametric uncertainty using generalized polynomial chaos," Computers and Chemical Engineering, vol. 76, pp. 63-75, 2015.

[6] D. Xiu, Numerical Methods for Stochastic Computation: A Spectral Method Approach, Princeton, New Jersey, USA: Princeton University Press, 2010.

[7] C. Chien, C. Hsu and P. Chen, "Semiconductor fault detection and classification for yield enhancement and manufacturing intelligence," Flex. Serv. and Manuf. J., vol. 25, no. 3, pp. 367-388, 2013.

[8] M. C. Keir and A. G. Alleyne, "Dynamic model-based fault detection and diagnosis residual considerations for vapor compression systems," in American Control Conference, 2006, Minneapolis, 2006.