How to Predict the Driving Range of a Lithium Ion Battery Using a Voltage Relaxation Curve
Most of the electronic devices that we use require batteries to power them. These batteries are now mostly lithium ion cells which have made it possible to pack more power in smaller sizes and lower cost. However, the battery life is limited by its internal degradation processes. The main method for determining the cycle life of Li-ion batteries is by measuring the depth of discharge (DoD) which is defined as the point at which a cell loses 80% of its original capacity.
When a cell is charged there are two phases: the initial charging process which is usually driven by a voltage regulator and the mass transfer phase that happens as ions move across the interface between the electrodes and electrolyte. The mass transfer process is influenced by the temperature of the electrodes and the electrolyte which in turn affects the chemistry of the cell. The chemistry of the cell can be described as an electrochemical reaction that requires energy for activation, concentration and polarization.
As the charge level of a battery drops during discharging the internal resistance of the cell increases and this is principally responsible for the rapid drop off of battery voltage near the end of a cycling test. In order to determine the cycling life of a battery it is therefore important to understand how the battery performs at different temperatures and discharging rates.
The ability to accurately identify the state of charge at a given depth of discharge enables the prediction of the driving range of an electric vehicle which is a primary concern for consumers. This article presents an approach to achieve this goal without the need for information from previous cycles using features extracted from a voltage relaxation curve. These features are converted to a model by applying transfer learning which can predict the state of charge with a root-mean-square error of 1.1%.
