Influence of Measurement and Prediction Uncertainties on Range Estimation for Electric Vehicles

Knowledge of the remaining driving range is crucial for drivers of battery electric vehicles (BEVs), because maximum range and charging infrastructure are limited. This and the fact that current implementations of range displays are deemed unreliable leads to range anxiety and ultimately less acceptance of BEVs. Every range estimation algorithm depends on various uncertain inputs and hence, its result is also uncertain. A more reliable range estimation could be achieved by assessing uncertainties of the underlying predictions and measurements and their influence on the calculation result. This paper presents a method for accurately calculating the probability of reaching the destination of a given route. For this, uncertainty of future velocity and battery state is analyzed using the example of an actual test vehicle. The error propagation through a physical model of the powertrain is evaluated by means of Monte-Carlo method and Taylor series expansion. Range uncertainty is investigated for different sets of input parameters, yielding a necessary safety margin between 12% and 23% battery charge. It is also shown that uncertainty more than doubles over lifetime of the vehicle and that neglecting this uncertainty can lead to unpredictable stranding for disadvantageous initial conditions. Knowledge of range uncertainty can reduce the necessary safety margin and avoid stranding of vehicles. In addition, displaying the uncertainty information to the driver alleviates range anxiety of drivers and consequently, acceptance of electric vehicles can be raised in society.