摘要：AbstractElectric race cars are a challenging application for battery management. The main issue is that the use of extremely high currents leads to additional nonlinear behaviour in the battery. The source of this nonlinear behaviour can be found in the nonlinear Butler-Volmer relation between currents and overpotentials, as well as self-heating that occurs when large currents are drawn due to the electrical resistance of the battery. As a result, nonlinearities in the input-output behaviour are caused by both factors. To accurately model the nonlinear overpotential behaviour using empirical battery models, it is necessary to be able to distinguish between the contribution of both sources of nonlinearities. In this paper, this problem is tackled by identifying a temperature- and current-dependent electrical model on lap data of an electric race car using a global approach to estimating state-space linear parameter-varying models. To aid the distinction between both effects, the influence of the temperature on the behaviour is distilled from local data, i.e., at constant temperatures. This is used as initialisation for the global optimisation problem, which identifies the effect of both phenomena from a single data set. Lap data of four race cycles is available. One cycle is used for parameter estimation of the battery model and the other three are used to validate the model. The results show that this approach brings a significant improvement to the modelling accuracy and presents opportunities to develop BMS applications, such as state estimators or even online power limiters for extreme battery-electric-vehicle applications.
