Power Demand Prediction Based on Mixed Driving Cycle Applied to Electric Vehicle Hybrid Energy Storage System

The use of multiple energy sources as power supply of an electric vehicle allows to improve its performance by increasing its autonomy and extending life cycle of on-board battery pack, which is the most expensive element of this type of automobile. In this work, it is proposed the use of computational intelligence techniques in the management of a hybrid energy storage system based on battery and supercapacitor, both embedded in an electric vehicle. For this purpose, a methodology for prediction and separation of the fractions of power demand, using an artificial neural network (ANN) based on the NonLinear Autoregressive Model with Exogenous Inputs (NARX), is presented. A mixed driving cycle (MDC), composed by the combination of characteristics of different standard driving cycles, is used for NARX ANN training. This proposed MDC ANN training allows a better performance of power demand prediction, when compared to a single driving cycle ANN training approach, since it imposes a more diversified speed profile. From the simulations carried out and the adjustments of the network parameters, a very small error was found in relation to the predicted signals. Based on the obtained results, it is possible to conclude that the proposed method is effective and promising for the calculation of power demand in electric vehicles.