A Prognostic and Filtering Analysis Technique for Estimating Battery Remaining Useful Life

Wednesday, 4 October 2017
Prince George's Exhibit Hall D/E (Gaylord National Resort and Convention Center)


Evaluating the life expectance of a lithium-ion battery is critical for cost analysis in small portable device development to large sustainable electric energy systems. Estimation of the serviceability of the battery/battery pack ensures users are effectively replacing legacy components with new ones at the most opportunistic moments, ensuring critical systems failure does not occur and that the essential functions are fully operational/supported. Prediction models have been developed to assist with the maintenance/upkeep of these devices.

Capacity fade modeling is capable of detecting the effects of the growth in the solid electrolyte interphase (SEI) layer through cycling within the battery cell. Capacity fade modeling techniques are helpful with identifying remaining useful life (RUL) and is a widely hailed techniques for battery prognostic. Miao developed an unscented particle filter capable of predicting RUL in [1] but didn’t take into account the effects of depth of discharge on the cyclability of the battery cell. Guo used cycling to monitor the loss of capacity and power fade for a battery in [2] but didn’t take into account the effects of temperature. Sustained usage at multiple temperatures can have an adverse effect on the nominal battery capacity during prolonged cycling. Development of a comprehensive RUL technique can be accomplished using filtering methods to simulate the behavior of a battery under the following criteria: temperature, multiple C-rate and depth of discharge (DoD).

RUL techniques used to estimate the serviceability of the battery cell define the capacity of the battery during all useable cycles compared to the capacity of a new cell multiplied by a degradation variable. The degradation variable is based on a Bayesian estimation process which considers temperature, C-rate, and DoD for multiple discharge scenarios. Fig 1 compares the filtering techniques supported by simulated data for multiple operating voltages at ambient temperature which enables better prognostics of the battery.


[1] Miao, Q., et al. “Remaing Useful Life Prediction of Lithium-Ion Battey With Unscented Particle Filter Technique.” Microelctronics Reliability 53 (2013) pp 805-810.

[2] Guo, J., et al. “A Bayesian Approach for Li-Ion Battery Capacity Fade Modeling and Cycles to Failure Prognostics.” Journal of Power Sources 281 (2015) pp 173-184.