Fault detection in serial connected high energy batteries – a statistical approach to estimate cell behavior without single cell voltage measurement.

摘要

The battery-management-system (BMS) in an electric vehicle has to assure safe operation and should lengthen battery life time. For most battery technologies, like nickel metal hydride, nickel cadmium, lead or inorganic lithium, single cell voltage measurement is not obligatory because safe operation is, to certain limits, not a question of overvoltage, overcharge or deep discharge. Therefore complexness of the BMS is reduced as far as possible in order to save costs. This results in systems measuring only a certain quantity of voltages and temperatures within the battery. Failures in electric vehicles using only the total battery voltage as reference have shown that this can not prevent serious damages. Further on more detailed voltage information looking at deep discharge or overcharge would be a nice feature in order to prevent unnecessary stress and to reduce aging processes. With the idea that in a good battery all cells should behave the same, it is possible to detect differences in their performance using the specific deviations. An assumed value for maximum and minimum cell voltage can be calculated. This helps to detect voltage breakdown, voltage rollover or overvoltage. Statistical calculations show that the number of voltage groups that has to be measured depends on the quality of the cells, in particular the variance of their open circuit voltages. A couple of requirements for the BMS was derived by investigating failures and applying the introduced method.