Nowadays, rechargeable Li-ion batteries have been widely used in laptops, cell phones and hybrid electric vehicles (HEV). The health information of battery is very important. In order to make diagnosis or prognosis, lots of work has been done. However, most of efforts are about the state of charge (SOC) estimation. For the battery capacity, which is a direct indicator of the battery health condition, there is still no well-established method. Most of the existing capacity estimation methods are not applicable mainly for three reasons: (1) difficulty and high expense for collecting data; (2) over confine of the battery working conditions; (3) long-term training.
In this thesis, after look into the knowledge and previous investigation methods of battery health, NASA battery data sets of 34 battery cells and IMS battery data sets of 8 battery cells are used for testing the proposed method. The student is trying to develop an applicable method for estimating the current capacity of a Li-ion battery without knowing any information of it except the type. Historical data of the same type of testing battery is needed for setting a historical data base. Then, with the data of current, time and temperature that technically collected from one charging process of the testing battery, an estimation of the current capacity of the testing battery can be made. Errors of the results from this method can be within the range of 20% with high probability. The advantages of this method are low cost, simple, quick, working condition and battery current state independent. With this method, people are able to get rough information about the battery health just by charging them.