Advanced

Driver Modeling, Velocity and Energy Consumption Prediction of Electric Vehicles

Bredberg, Carl and Stjernrup, John (2017)
Department of Automatic Control
Abstract
A driver model can be used to predict the vehicle velocity and the energy consumption. It can be modeled such that it benefits from historical data and can be further improved if a navigation system is available. A well implemented driver model is important since the car fleet seems to be more and more electrified. The role of the driver model would then be to increase the accuracy for when the driver needs to take a break to recharge the vehicle, and thereby decrease the driver’s range anxiety.
Historical behaviour of different drivers has been measured and collected by Volvo Car Corporation. The information regarding these drivers has been used in four out of five implemented driver models. Three of the models use Markov chain theory... (More)
A driver model can be used to predict the vehicle velocity and the energy consumption. It can be modeled such that it benefits from historical data and can be further improved if a navigation system is available. A well implemented driver model is important since the car fleet seems to be more and more electrified. The role of the driver model would then be to increase the accuracy for when the driver needs to take a break to recharge the vehicle, and thereby decrease the driver’s range anxiety.
Historical behaviour of different drivers has been measured and collected by Volvo Car Corporation. The information regarding these drivers has been used in four out of five implemented driver models. Three of the models use Markov chain theory to make the prediction while the fourth takes advantage of frequency analysis. Above the aim to increase the accuracy of the energy consumption prediction it is investigated to what extent a personal driver model can be created.

In addition to the driver models a primitive method to predict when a driver reacts to a new posted reference speed is proposed and four validation methods are suggested.
The results indicate that the driver models based on historical data perform better energy predictions than the one without any historical data. The driver model that uses Gaussian mixture model together with Markov chain theory makes the best energy prediction. The individual differences are especially shown at road segments with higher reference speed. In urban traffic it is more likely that the traffic pattern decides the energy consumption. (Less)
Please use this url to cite or link to this publication:
author
Bredberg, Carl and Stjernrup, John
supervisor
organization
year
type
H3 - Professional qualifications (4 Years - )
subject
report number
TFRT-6035
ISSN
0280-5316
language
English
id
8923444
date added to LUP
2017-09-08 13:19:17
date last changed
2017-09-08 13:19:17
@misc{8923444,
  abstract     = {A driver model can be used to predict the vehicle velocity and the energy consumption. It can be modeled such that it benefits from historical data and can be further improved if a navigation system is available. A well implemented driver model is important since the car fleet seems to be more and more electrified. The role of the driver model would then be to increase the accuracy for when the driver needs to take a break to recharge the vehicle, and thereby decrease the driver’s range anxiety.
 Historical behaviour of different drivers has been measured and collected by Volvo Car Corporation. The information regarding these drivers has been used in four out of five implemented driver models. Three of the models use Markov chain theory to make the prediction while the fourth takes advantage of frequency analysis. Above the aim to increase the accuracy of the energy consumption prediction it is investigated to what extent a personal driver model can be created.

 In addition to the driver models a primitive method to predict when a driver reacts to a new posted reference speed is proposed and four validation methods are suggested.
 The results indicate that the driver models based on historical data perform better energy predictions than the one without any historical data. The driver model that uses Gaussian mixture model together with Markov chain theory makes the best energy prediction. The individual differences are especially shown at road segments with higher reference speed. In urban traffic it is more likely that the traffic pattern decides the energy consumption.},
  author       = {Bredberg, Carl and Stjernrup, John},
  issn         = {0280-5316},
  language     = {eng},
  note         = {Student Paper},
  title        = {Driver Modeling, Velocity and Energy Consumption Prediction of Electric Vehicles},
  year         = {2017},
}