A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies

Paul Karoshi, Karin Tieber, Christopher Kneissl, Georg Peneder, Harald Kraus, Martin Hofstetter, Jürgen Fabian, Martin Ackerl

Publikation: Beitrag in Buch/Bericht/KonferenzbandBeitrag in einem KonferenzbandForschungBegutachtung

Abstract

In hybrid electric vehicles (HEV), the operation strategy strongly influences the available system power, as well as local exhaust emissions. Predictive operation strategies rely on knowledge of future traction-force demands. This predicted information can be used to balance the battery’s state of charge or the engine’s thermal system in their legal operation limits and can reduce peak loads. Assuming the air and rolling drag-coefficient to be constant, the desired vehicle velocity, vehicle-mass and longitudinal driving resistances determine the vehicle’s traction-force demand. In this paper, a novel methodology, combining a history-based prediction algorithm for estimating future traction-force demands with the parameter identification of road grade angle and vehicle mass, is proposed. It is solely based on a route-history database and internal vehicle data, available on its on-board communication and measuring systems. It complements state-of-the-art navigation software, as these systems usually are not activated on frequently driven routes. In a first step, a Kalman filter estimates the vehicle mass and the current grade angle of the road online, using the vehicle’s longitudinal equation of motion. In a second step, velocity and road gradient are predicted. This is done by comparing online vehicle data with data stored in a route history. As the steering-wheel angle correlates well with the position on a given route, it is chosen as distinctive parameter for route identification. A longitudinal vehicle model calculates the approximated future traction-force demand from the predicted velocity and road gradient trajectory, considering the online estimated vehicle mass. Then the operation strategy can determine control variables, such as the upcoming loads to the propulsion units, for a certain prediction horizon ahead of the vehicle. Validation results of the prediction system are presented for an all-electric passenger car. However, computing and memory requirements for a real-time capable hardware are not considered.
Originalspracheenglisch
TitelSAE - World Congress 2016
Herausgeber (Verlag)SAE International
PublikationsstatusVeröffentlicht - 14 Apr 2016
VeranstaltungSAE World Congress 2016 - Detroit, USA / Vereinigte Staaten
Dauer: 12 Apr 201614 Apr 2016

Konferenz

KonferenzSAE World Congress 2016
LandUSA / Vereinigte Staaten
OrtDetroit
Zeitraum12/04/1614/04/16

Schlagwörter

  • Predictive operation strategy
  • road grade estimation
  • mass estimation
  • real time capability

Dies zitieren

Karoshi, P., Tieber, K., Kneissl, C., Peneder, G., Kraus, H., Hofstetter, M., ... Ackerl, M. (2016). A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies. in SAE - World Congress 2016 [2016-01-1238] SAE International.

A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies. / Karoshi, Paul; Tieber, Karin; Kneissl, Christopher ; Peneder, Georg; Kraus, Harald; Hofstetter, Martin; Fabian, Jürgen; Ackerl, Martin.

SAE - World Congress 2016. SAE International, 2016. 2016-01-1238.

Publikation: Beitrag in Buch/Bericht/KonferenzbandBeitrag in einem KonferenzbandForschungBegutachtung

Karoshi, P, Tieber, K, Kneissl, C, Peneder, G, Kraus, H, Hofstetter, M, Fabian, J & Ackerl, M 2016, A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies. in SAE - World Congress 2016., 2016-01-1238, SAE International, SAE World Congress 2016, Detroit, USA / Vereinigte Staaten, 12/04/16.
Karoshi P, Tieber K, Kneissl C, Peneder G, Kraus H, Hofstetter M et al. A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies. in SAE - World Congress 2016. SAE International. 2016. 2016-01-1238
Karoshi, Paul ; Tieber, Karin ; Kneissl, Christopher ; Peneder, Georg ; Kraus, Harald ; Hofstetter, Martin ; Fabian, Jürgen ; Ackerl, Martin. / A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies. SAE - World Congress 2016. SAE International, 2016.
@inproceedings{11278ba45f934b1f8a8fecff80254867,
title = "A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies",
abstract = "In hybrid electric vehicles (HEV), the operation strategy strongly influences the available system power, as well as local exhaust emissions. Predictive operation strategies rely on knowledge of future traction-force demands. This predicted information can be used to balance the battery’s state of charge or the engine’s thermal system in their legal operation limits and can reduce peak loads. Assuming the air and rolling drag-coefficient to be constant, the desired vehicle velocity, vehicle-mass and longitudinal driving resistances determine the vehicle’s traction-force demand. In this paper, a novel methodology, combining a history-based prediction algorithm for estimating future traction-force demands with the parameter identification of road grade angle and vehicle mass, is proposed. It is solely based on a route-history database and internal vehicle data, available on its on-board communication and measuring systems. It complements state-of-the-art navigation software, as these systems usually are not activated on frequently driven routes. In a first step, a Kalman filter estimates the vehicle mass and the current grade angle of the road online, using the vehicle’s longitudinal equation of motion. In a second step, velocity and road gradient are predicted. This is done by comparing online vehicle data with data stored in a route history. As the steering-wheel angle correlates well with the position on a given route, it is chosen as distinctive parameter for route identification. A longitudinal vehicle model calculates the approximated future traction-force demand from the predicted velocity and road gradient trajectory, considering the online estimated vehicle mass. Then the operation strategy can determine control variables, such as the upcoming loads to the propulsion units, for a certain prediction horizon ahead of the vehicle. Validation results of the prediction system are presented for an all-electric passenger car. However, computing and memory requirements for a real-time capable hardware are not considered.",
keywords = "Predictive operation strategy, road grade estimation, mass estimation, real time capability, Predictive operation strategy, mass estimation, road grade estimation, real time capability, commercial vehicle",
author = "Paul Karoshi and Karin Tieber and Christopher Kneissl and Georg Peneder and Harald Kraus and Martin Hofstetter and J{\"u}rgen Fabian and Martin Ackerl",
year = "2016",
month = "4",
day = "14",
language = "English",
booktitle = "SAE - World Congress 2016",
publisher = "SAE International",
address = "United States",

}

TY - GEN

T1 - A History-Based Load Requirement Prediction Algorithm, for Predictive Hybrid- and Thermal Operation Strategies

AU - Karoshi, Paul

AU - Tieber, Karin

AU - Kneissl, Christopher

AU - Peneder, Georg

AU - Kraus, Harald

AU - Hofstetter, Martin

AU - Fabian, Jürgen

AU - Ackerl, Martin

PY - 2016/4/14

Y1 - 2016/4/14

N2 - In hybrid electric vehicles (HEV), the operation strategy strongly influences the available system power, as well as local exhaust emissions. Predictive operation strategies rely on knowledge of future traction-force demands. This predicted information can be used to balance the battery’s state of charge or the engine’s thermal system in their legal operation limits and can reduce peak loads. Assuming the air and rolling drag-coefficient to be constant, the desired vehicle velocity, vehicle-mass and longitudinal driving resistances determine the vehicle’s traction-force demand. In this paper, a novel methodology, combining a history-based prediction algorithm for estimating future traction-force demands with the parameter identification of road grade angle and vehicle mass, is proposed. It is solely based on a route-history database and internal vehicle data, available on its on-board communication and measuring systems. It complements state-of-the-art navigation software, as these systems usually are not activated on frequently driven routes. In a first step, a Kalman filter estimates the vehicle mass and the current grade angle of the road online, using the vehicle’s longitudinal equation of motion. In a second step, velocity and road gradient are predicted. This is done by comparing online vehicle data with data stored in a route history. As the steering-wheel angle correlates well with the position on a given route, it is chosen as distinctive parameter for route identification. A longitudinal vehicle model calculates the approximated future traction-force demand from the predicted velocity and road gradient trajectory, considering the online estimated vehicle mass. Then the operation strategy can determine control variables, such as the upcoming loads to the propulsion units, for a certain prediction horizon ahead of the vehicle. Validation results of the prediction system are presented for an all-electric passenger car. However, computing and memory requirements for a real-time capable hardware are not considered.

AB - In hybrid electric vehicles (HEV), the operation strategy strongly influences the available system power, as well as local exhaust emissions. Predictive operation strategies rely on knowledge of future traction-force demands. This predicted information can be used to balance the battery’s state of charge or the engine’s thermal system in their legal operation limits and can reduce peak loads. Assuming the air and rolling drag-coefficient to be constant, the desired vehicle velocity, vehicle-mass and longitudinal driving resistances determine the vehicle’s traction-force demand. In this paper, a novel methodology, combining a history-based prediction algorithm for estimating future traction-force demands with the parameter identification of road grade angle and vehicle mass, is proposed. It is solely based on a route-history database and internal vehicle data, available on its on-board communication and measuring systems. It complements state-of-the-art navigation software, as these systems usually are not activated on frequently driven routes. In a first step, a Kalman filter estimates the vehicle mass and the current grade angle of the road online, using the vehicle’s longitudinal equation of motion. In a second step, velocity and road gradient are predicted. This is done by comparing online vehicle data with data stored in a route history. As the steering-wheel angle correlates well with the position on a given route, it is chosen as distinctive parameter for route identification. A longitudinal vehicle model calculates the approximated future traction-force demand from the predicted velocity and road gradient trajectory, considering the online estimated vehicle mass. Then the operation strategy can determine control variables, such as the upcoming loads to the propulsion units, for a certain prediction horizon ahead of the vehicle. Validation results of the prediction system are presented for an all-electric passenger car. However, computing and memory requirements for a real-time capable hardware are not considered.

KW - Predictive operation strategy

KW - road grade estimation

KW - mass estimation

KW - real time capability

KW - Predictive operation strategy

KW - mass estimation

KW - road grade estimation

KW - real time capability

KW - commercial vehicle

M3 - Conference contribution

BT - SAE - World Congress 2016

PB - SAE International

ER -