In order to accurately estimate the remaining capacity of a battery, it is critical that the full capacity of the battery be accurately known. Unfortunately, under normal use conditions such as those encountered in an electric vehicle or other battery-power device, it is difficult to accurately ascertain battery capacity. For example, in one method of determining battery capacity, the initial capacity of the battery is gradually decreased based on a variety of factors such as battery age, the number of charge/discharge cycles to date, and temperature. Unfortunately this technique does not provide a very accurate assessment of battery capacity, both because some factors are not properly taken into account (e.g., historical temperature profiles, load conditions, depth of discharge prior to each charging, charge/discharge rates, etc.) and because the effects of the errors accumulate as the battery ages. Another method of determining battery capacity is to allow the battery to become fully discharged, and then determine the capacity of the battery during charging. Although this technique can be used occasionally, using it on a routine basis can have serious repercussions since deep discharging a battery, and in particular fully discharging a battery, can dramatically shorten its lifetime. Additionally, for most battery-powered devices, especially electric vehicles, it would be extremely inconvenient to require that the user allow the battery to become fully discharged prior to charging. This would be similar to requiring that a conventional car be driven until the gas tank was dry before refilling, simply in order to determine the gas tank's capacity.
...
what is needed is a method of accurately estimating battery capacity that is less susceptible to, or more accurately takes into account, the many factors that can influence a battery's capacity.
...
[0005]The present invention provides a method and apparatus for accurately estimating battery capacity based on a weighting function. The disclosed system monitors battery current and uses the monitored battery current to calculate the state of charge (SOCbyAh) of the battery.
...
weighting calculation are based on sensor accuracies and control loop timing. These constants, like many non-linear real-world systems, are difficult to model and therefore are generally tuned based on real-world experimental data, the intended use of the device, and how much overshoot and response time can be tolerated. Typically, the initial settings for the constants are based on sensor accuracy and usage assumptions. For instance, in an EV application where the consequences of overshoot are severe and there are frequent rests (e.g., when the driver stops at the grocery store, school, restaurant, etc.), these constants are set high to give a low weighting.
...