Online capacity estimation of lithium-ion batteries with deep long short-term memory networks

There is an increasing demand for modern diagnostic systems for batteries under real-world operation, specifically for the estimation of their state of health, for example, via their remaining capacity. The online estimation of the capacity of a cell is challenging due to the dynamic nature of cell aging and the limited variety of inputs available from a cell under operation. The scope of this work is the development of a data-driven capacity estimation model for cells under real-world working conditions with recurrent neural networks having long short-term memory capability. Voltage-time sensor data from the partial constant current phase charging curve is used as input, reflecting input availability in the real world. The network achieves a best-case mean absolute percentage error of 0.76% and is extremely robust while handling input noise. It also has the ability to handle variations in the length of the input time series and can generate a viable estimation even with an incomplete collection of input due to sensor errors. The model validation with several scenarios is done in a local embedded device, highlighting the use case of such models in future battery management systems.

Automated summary

There is a growing demand for modern diagnostic systems for batteries in real-world operation, especially for estimating their health status such as remaining capacity. A data-driven capacity estimation model using recurrent neural networks with long short-term memory capability has been developed for cells under real-world working conditions. This model is robust, can handle input noise, adapt to variations in input time series length, and generate viable estimation even with incomplete input data.