A system for cloud-based deviation prediction of propulsion energy consumption for EVs

Energy prediction for electric vehicles (EVs) is a complex problem because the energy consumption depends on a lot of different and varying impact factors. Since the number of vehicles connected to a server will increase, cloud-based approaches can improve the accuracy of energy prediction for EVs. A prediction model is used, which consists of an in-vehicle part for the prediction of the mean value of propulsion energy consumption and of a cloud-based part to predict the relative deviation from a normalized mean energy consumption value on the basis of collected speed profiles. In this paper, the cloud-based part for the deviation prediction is introduced, which can be used for EVs with different vehicle attributes. Extracted statistical features from collected speed profiles, which are stored on a server in the backend, are used as input for multiple regression prediction models. Variations in speed profiles, which can be caused by individual driving behaviour for example, can be considered with the prediction model.