Sunteți pe pagina 1din 15

Predictive Operation Strategy for Hybrid

Vehicles

M. Sc. Jue Wang


Prof. Dr.-Ing. Hermann Koch-Groeber

Heilbronn University
Faculty of Mechanics and Electronics
Automotive Systems Engineering

1
Predictive Operation Strategy for Hybrid Vehicles

1 Introduction
The hybrid electric vehicle (HEV) is considered to be one of the best solutions for the
automobile industry, to cope with the diminishing oil resources and environmental
problems, by achieving low emissions and fuel consumption. The HEV is powered by
two sources, composed of an internal combustion engine (ICE) and its transmission as
well as one or two electric machines (EM), battery pack and converters. The
additional electric power system brings further degrees of freedom for powertrain
arrangement and operation, which lead to a challenge the research and development.
A number of publications [Radke13] [Schroeter13] [Albers13] [Ambhl09]
[Stiegeler08] discussed predictive operation strategies with global optimization
methods, e.g. dynamic programming. However, the computing and memory
requirements of the global optimization method are significantly higher than those
from a heuristic optimization method, thus rendering its complete application in on-
board electronic control units (ECU) almost impossible. Some predictive operation
strategies are combinations of global optimization and heuristic optimization method
[Katsargyri09] [LaSch13]. In order to develop an online operation strategy for HEV
by using the traffic context for semi-automated driving, a heuristic method is applied
in this work and with consideration of longitudinal dynamics. Beside the fuel
consumption the driving comfort and operation complexity are considered in this
method to be optimized, which is lacked in many publications. This work presents the
method and simulation results of this strategy. The details about real time capability
will be done in the future work.

2 Predictive Operation Strategy


The goal of this predictive operation strategy is an online optimization of fuel
consumption and powertrain operation by using the information of future route for
full-hybrid as well as plug-in hybrid vehicles (PHEV). The heuristic operation
strategy of this work is derived from an optimized solution by integrating the longtime
driving prediction functionality. This integration is realized by the estimation of future
operating modes and the SOC changes by a regressive analysis and criteria with
regard to the future route.

2.1 Short - and Long-Distance Zone and Speed Profile Prediction


The vehicle speed prediction is estimated by using the traffic and road information
which are obtained from a navigation (Navi) system, adaptive cruise control (ACC)
system and other sensors e.g. radars of the vehicle. Due to the calculation cost and
different ranges covered by the sensors two zones as shown in Figure1, short- and

2
Predictive Operation Strategy for Hybrid Vehicles

long-distance zone, are defined to estimate the vehicle speed and determinate the
operation modes of HEV. The main task of the short-distance zone (approx. 200
meter) is to derive an accurate current operation mode by the typical detection range
of sensors applied to todays most common drive assistance systems [Wall06]. The
range of the long-distance zone should be a value which balance the quality of the
optimization and calculation cost. As a first approach a distance, which corresponds a
duration of 20 minutes, is defined as the range of the long-distance zone. This range is
the usual pure electrical range of common PHEVs, approx. 50 km, considering the
highway driving with average speed 130 km/h. In that case, the complete electrical
energy of the battery is taken into account in the optimization and driving with empty
battery can be avoided.

Figure1. Vehicle speed prediction in the short- and long-distance zone

The vehicle speed profile is estimated with considerations of driving style of the
driver, road speed limits and other traffic information, such as vehicle speed of a car
which is driving in front of the HEV. The speed limits of the future route will be
derived from the navigation system, as well as from speed limit signs recognized by a
camera based system. In addition, the traffic information is considered to be another
boundary condition of the speed profile. The driving style of the driver decides the
accurate cruise speed, acceleration and deceleration. Furthermore the previous
research [KoGr12] [KoGr13] [KGWa14] show, that HEVs also benefit from the
coasting. Thus the integration of coasting phases in the speed profile is derived by
using the road grade information of the future route. The calculation effort for the
long distance zone remains small due to a limited number of steps (e.g. max. 40),

3
Predictive Operation Strategy for Hybrid Vehicles

defined by relevant changes of driving conditions, for instance speed limits, peak and
low points of road profile or congested traffic.

2.2 Regressive Determination of the Operation Mode


After the speed profile prediction the operation modes are determined regressively
first in the long-distance zone and then in the short-distance zone, the later in a higher
calculation frequency. Figure2 shows the complete procedure of the method. The
update time for predictive speed profile and operation modes determination in the
long-distance zone are the same (TL). The update time for operation modes
determination in the short-distance zone TS is approx. 10 times smaller than TL.

Figure2. Procedure of the predictive operation strategy

The same algorithm of regressive determination is applied in both zones. Figure3


shows the algorithm in the long-distance zone with help of an example vehicle speed
profile.

Figure3. Algorithm for determination of the operation mode in long-distance zone

4
Predictive Operation Strategy for Hybrid Vehicles

The long-distance zone is discretized with n points. The mode at point 1, which is at
the end of the long-distance zone, is set to be an ideal mode, e.g. pure electric driving.
Mode candidates on point 2 are selected according to the driving speed profile and
vehicle status. Each candidate is evaluated with a note, as shown Formula1, which
includes three sub-notes -- efficiency, mode change complexity and SOC-leveling --
and weight factors w1,2,3. The candidate with the highest note is chosen as operation
mode. This evaluation and selection method is applied step by step till the point n,
which is at the beginning of the long-distance zone. Similar to the procedure in the
long-distance zone the operation modes in the short-distance zone are determined.
= 1 + 2 + 3 (1)
The mode efficiency sub-note Noteefficiency is defined as the tank-to-wheel efficiency,
see Equation2.
= /( + ) (2)
, where the Edmd is the demanded energy by the vehicle, the Etank is the fuel
consumption from the tank and the Ebat is the electric consumption of the battery.
The sub-note Notecompelexity takes the consideration of the mode change complexity
between the previous operation mode and the current mode. For example, if the
previous mode is pure electric driving, the ICE-solo mode at current point has a lower
note than the pure electric driving mode, due to the on/off switch of the ICE, see
Table1. This evaluation is used to avoid frequently switching of the ICE as well as
EM and uncomfortable vibration caused by the mode change.

Table1. Mode change complexity note from EM solo


ICE-solo Hybrid EM-solo
EM-solo 1 2 3

As the full hybrid vehicle the charge strategy is charge sustaining method, which
means the SOC of the battery pack should be hold at a certain level, e.g. between 60%
and 70%. The sub-note NoteSOC-Leveling enables precautionary charging and discharging
of the battery according to the future drive operating mode, which is derived by the
SOC change in the future steps. For example, before deceleration phases, the vehicle
will enable pure electric driving to leave enough capacity for the recuperation energy,
shown in Table2.

5
Predictive Operation Strategy for Hybrid Vehicles

Table2. Mode SOC-Leveling note before deceleration


Conditions Note for Current Mode
SOC1 SOC2 ICE-solo Hybrid EM-solo
>0 >0 2 1 3

SOC1 and SOC2 are SOC changes in the following two calculation points of the
future route, which are positive due to the recuperation in the next two sections. As
defined in Equation3 and Equation4, Qb is the nominal capacity of the battery; Q1
and Q2 are electric consumptions in the next two calculation section. For example,
the current point is LZ4 in Figure3 and SOC1 is SOCLZ2 minus SOCLZ3, SOC2 is
SOCLZ1 minus SOCLZ2.
1 = 1 / (3)
2 = 2 / (4)
As mentioned before, the main goal of the long-distance zone is the SOC balancing,
therefore, this sub-note NoteSOC-Leveling will be prioritized with help of the weight
factor w3 in the operation strategy in the long-distance zone. In contrast, the operation
strategy in the short-distance zone focuses more on the driving comfort, and therefore
provides a different set of weight factors, which emphasizes the Notecomplexity with a
higher w2 value.

3 Modeling and Simulation Results


Modeling and simulation calculations in this study are performed by the commercial
software CRUISE of AVL List GmbH and Matlab/Simulink, in order to
demonstrate the performance of the previous described predictive operation strategy.
For that case a preliminary strategy for a real highway driving condition is simulated
herein.

3.1 Hybridization Modeling


The reference vehicle of the research project ECC [KoGr13] is hybridized virtually
and used in this work. The simulation results of this contribution are basing on a car
of the same type: Ford (of Europe) Focus Sedan 5-door with 1.6 T ecoboost 110 kW
6-speed MT model year 2012. It is equipped with a modern turbo-charged 4 cylinder
gasoline DI-engine as moderate downsizing concept. The characteristics of the main
powertrain components are given in Table3.

6
Predictive Operation Strategy for Hybrid Vehicles

Table3. Data of reference vehicle


Parameter Value
1.6 T ecoboost Max. Power 110 kW @ 5700 min-1
gasoline DI-engine Max. Torque 240 Nm from 1600 to 4000 min-1 constant
Displacement 1596 cm
Mass 1362 kg (including driver and fuel)
Drag 0.678 m (cw x Across.sec: 0.30 x 2.26 m)
Air density 1.2 kg/m
Rolling friction factor 0.01 independent of velocity
Total efficiency powertrain 0.95
Rotational masses factor lambda 1.03
Top Speed 210 km/h
Fuel Consumption NEDC 5.9 l/100km (combined: 137 g CO2/km)
7.6 l/100km (urban)
4.9 l/100km (extra-urban)
Idle fuel consumption 0.6 l/h
Fuel to restart engine 0.4 ml
Asynchron Max. rotational speed 7500 min-1
ous Max. Torque 240 Nm @ 3000 min-1
Machine Max. Power 75 kW
Battery Type Lithium Iron Phosphate
Nominal capacity 40 Ah
(cell)
Nominal voltage 252 V
(pack)
Usable energy (pack) approx. 10 kWh

The reference vehicle Ford Focus is virtually hybridized by adding the electrical drive
axle on the rear axle, see Figure4. The pure electrical range of this hybridized vehicle
with an approx. 10 kWh battery pack is about 54 km for constant 130 km/h highway
driving, which is typical pure electrical range for PHEVs e.g. Audi A3. The electric
machine E-Motor/Generator is directly connected with the rear axle, which can
either deliver the traction power or charge the battery by braking energy recuperation.
This axle split hybrid concept is easier to implement due to less changes of the
arrangement, e.g. no additional changes for the conventional front drive axle. The
direct connection of the electric machine has relative high efficiency during
recuperation. Therefor this concept is applied in this work. The main components

7
Predictive Operation Strategy for Hybrid Vehicles

such as ICE and EM are modelled based on efficiency maps. The forward-looking
powertrain and operation strategy model is built in Matlab/Simulink.

Wheel Wheel

AC/DC
Differential Gearbox Clutch ICE Battery EM/G Differential
DC/AC

Wheel Wheel

front axle rear axle

mechanical connections
electrical connections

Figure4. Hybridized Ford Focus axle split hybrid powertrain configuration

Five operation modes for this HEV are defined:


1 ICE-solo: pure conventional driving
2 Hybrid: driving by both ICE and EM, boost or battery recharging mode
activation based on drivetrain operation
3 EM-solo: pure electric driving
4 Recuperation: energy is recuperated by EM while deceleration
5 Coasting: rolling with declutched drivetrain and ICE turned off

3.2 Reference Road Profiles


The reference road is a real profile from hilly German motorway A81 between exits
Pleidelsheim and Ilsfeld, totally 11.4 km, basing on data from the LGL (office for
geoinformation of the state of Baden-Wrttemberg). This three lane motorway had
been completely rebuilt in the 1980s, such that its profile may be taken as a sample
applying modern highway tracing. Figure5 shows the altitude and corresponding
grade profile. The altitude data is filtered by the Gaussian filter.

8
Predictive Operation Strategy for Hybrid Vehicles

Figure5. Road profile of A81 Pleidelsheim-Ilsfeld

3.3 Vehicle Speed Profile Prediction


According to observed speed limits, there is no speed limitation for the first 9.6 km,
and further route with limitation 100 km/h. As described in section 2.1, the vehicle
speed profile is predicted by this speed limit of the reference motorway. The vehicle is
driving normally with constant speed, "recommended speed" 130 km/h for road
section without limit and 100 km/h for the section with speed limitation. Before the
limited road section, the vehicle decelerates from 130 km/h to 100 km/h with the
deceleration of -1 m/s2. By this deceleration the battery is charged with its maximal
charge power approx. 30 kW as well as charge current approx. 180 A. The first
predicted vehicle speed profile v0 is shown in Figure6.

Figure6. First prediction of the vehicle speed v0

9
Predictive Operation Strategy for Hybrid Vehicles

In order to reduce the fuel consumption especially on the hilly motorway, the coasting
functionality is integrated in the predicted speed profile. The integration of coasting
according to the following principles:
The coasting phase begins at the point, where the grade is negative.
The vehicle is allowed to coasting till the low speed limit is reached, 10%
less than predefined cruise speed.
The vehicle accelerates again with 1m/s2 after coasting, till the predefined
cruise speed.
Figure7 shows the final prediction of the speed profile v. The downhill grade is too
steep for coasting (from 9.6 km) since the cars acceleration would make it exceed the
speed limit (100 km/h). Therefor the last coasting phase begins relative late and does
not completely locate in the downhill section.

Figure7. Final prediction of the vehicle speed v

3.4 Simulation Results of the regressive predictive Operation


Strategy
The regressive operation strategy is simulated with a predictive vehicle speed profile.
Figure8 shows the operation modes are determined according to calculation in the
long-distance zone, which is the total route. It is obvious that with this regressive
method the SOC of the battery is under precautionary control. Before the deceleration
the vehicle is driving purely electric to gain capacity for the energy which will be
recuperated later. Furthermore before long pure electric driving (from approx.1.2 to
2.7 km) the battery is charged earlier (from 0 to approx. 0.9 km).

10
Predictive Operation Strategy for Hybrid Vehicles

310
290
altitude (m)

270
250
230
210
0 2 4 6 8 10 12

150 5 Coasting

130 4 Recuperation

Operation mode
v (km/h)

110 3 EM- solo

90 2 Hybrid

70 1 ICE-solo
0 2 4 6 8 10 12
distance (km) v
operation mode

Figure8. Determined operation modes of calculation in long-distance zone

The next step of the method is operation modes determination in the short-distance
zone based on the results of determination in the long-distance zone. As mentioned in
Section2.2 the goal in the short-distance zone is driving comfort, which is different
from the goal in the long-distance zone. As it is shown in Figure9 comparing with the
result only the long-distance zone, the additional consideration of short-distance
reduces unnecessary EM on/off switch (at approx. 3 km and 5.7 km). Before the
Coasting driving the ICE is switched off earlier to save fuel. After the determination
in the short-distance zone the final SOC of battery is identical to only long-distance
zone calculation.

11
Predictive Operation Strategy for Hybrid Vehicles

150 5 Coasting

Operation mode
130 4 Recuperation
v (km/h)

3 EM-solo
110

90 2 Hybrid

70 1 ICE-solo
0 2 4 6 8 10 12
68
SOC (%)

65
v (km/h)
62 Mode LZ
Mode LZ+SZ
59
0 2 4 6 8 10 12
distance (km)

Figure9. Comparison of SOC and operation modes as the results in long-distance zone (LZ) and
in both zones (LZ+SZ)

As Table4 shown the fuel consumption for the final determined operation of the
predictive strategy is 5.99 L/100km, which is 0.24 L/100km more than the result of
calculation in only long-distance zone, due to the consideration of mode change
complexity.

Table4. Fuel consumptions and electric consumptions of different strategies


Fuel consumption ele. consumption
(L/100km) (kWh)
predictive LZ 5.75 0.06
predictive LZ+SZ 5.99 0.08
Rule based 6.89 0.2
Rule based (the same vmean) 6.73 0.06

In order to evaluate the quality of the predictive operation strategy in this work, a
simple rule based operation strategy is simulated in Matlab/Simulink and AVL Cruise
with the same route and HEV powertrain concept. The operation mode decisions
depend on the instantaneous driving speed, acceleration and SOC of the battery
[Salmasi07]. As Figure10 shown the vehicle is driving with rule based strategy, which
is without coasting sections. Comparing with this rule based strategy there is 13% less
fuel consumption given by the predictive operation strategy. This saving potential is
partially benefited from the reduced average speed vmean caused by the coasting.
Therefore the rule based strategy with the same vmean on the same route is simulated,

12
Predictive Operation Strategy for Hybrid Vehicles

which fuel consumption is 6.67 L/100km. The saving potential of the regressive
predictive strategy is reduced to approx.10% by comparing the simulation with the
same vmean.

Figure10. Comparison of rule based strategy and predictive operation strategy

4 Conclusions
A new method for the optimization of a semi-automated HEV was developed. This
method consist two main steps, vehicle speed profile prediction and regressive
operation strategy. The functionalities of this method are:
The long-distance zone enables a long time prediction method, which use the
information of navigation and positioning systems. The short-distance zone
enables the predictive operation, which is affected by the traffic context.
Fuel consumption is reduced by the integration of coasting functionality.
Unnecessary mode change and ICE and EM switch are avoided by the
evaluation of the mode change complexity.
Precautionary SOC control is realized by the evaluation of the SOC balance
according to future route.
In this work the fuel saving potential of the new method is up to 10% comparing with
rule based strategy.

13
Predictive Operation Strategy for Hybrid Vehicles

5 Acknowledgements
The author thanks for the funding of this research project ECC economic cruise
control in the program IngenieurNachwuchs 2010 (young engineering talents) of
BMBF and for the support by the project partners:
IPEK Institut fr Produktentwicklung at Karlsruhe Institute of Technology
(KIT), sincerest thanks to o. Prof. Dr.-Ing. Dr. h.c. A. Albers, Dr.-Ing. Matthias
Behrendt and all colleagues in the institute.
GETRAG Getriebe- und Zahnradfabrik Hermann Hagenmeyer GmbH & Cie KG,
D-74199 Untergruppenbach
IPETRONIK GmbH & Co. KG, D-76532 Baden-Baden
RA Consulting GmbH, D-76646 Bruchsal

Literature
[KoGr12] Koch-Groeber H., Niermann H., Spoerer S., Vogelmann D.,
Wittemann E., ECC Economic Cruise Control, VPC Virtual
Powertrain Creation Conference, Esslingen, 2012
[KoGr13] Koch-Groeber H., Coasting on Highways Potentials and
Realization; 13th Stuttgart Symposium, 2013
[KGWa14] Koch-Groeber H., Wang J., Criteria for Coasting on Highways for
Passenger Cars ; No. 2014-01-1157.SAE Technical Paper, 2014
[Radke13] Tobias Radke, Energieoptimale Lngsfhrung von Kraftfahrzeugen
durch Einsatz vorausschauender Fahrstrategien, Dissertation, 2013,
Karlsruher Institut fr Technologie (KIT), Karlsruhe
[Schroeter13] Jens Schrter, Das erweiterte X-in-the-Loop-Framework zur
durchgngigen Integration von Optimierungs-verfahren in den
Produktentwicklungsprozess am Beispiel der Entwicklung
energieeffizienter Fahrzeuge, Dissertation, 2013, Karlsruher Institut
fr Technologie (KIT), Karlsruhe
[Albers13] Albers, A., Behrendt, M., Schroeter, J., Ott, S., & Klingler, S. (2013).
X-in-the-Loop: A Framework for Supporting Central Engineering
Activities and Contracting Complexity in Product Engineering
Processes. In Proceedings of the International Conference on
Engineering Design, ICED 2013

14
Predictive Operation Strategy for Hybrid Vehicles

[Ambhl09] Daniel Ambhl, Energy Management Strategies for Hybrid Electric


Vehicles, Dissertation, 2009, ETH Zrich
[Stiegeler08] Markus Stiegeler, Entwurf einer vorausschauenden Betriebsstrategie
fr parallele hybride Antriebsstrnge, Dissertation, 2008,
Universitt Ulm
[Katsargyri09] G.-E. Katsargyri, et al. Path dependent receding horizon control
policies for Hybrid Electric Vehicles. Control Applications, (CCA)
& Intelligent Control, (ISIC), 2009 IEEE. 2009. 607-612.
[LaSch13] Stephan Lange, Michell Schimanski Energiemanagement in
Fahrzeugen mit alternativen Antrieben, Dissertation, 2006,
TU Braunschweig, Braunschweig
[Wall06] H. Wallentowitz, ed., Handbuch Kraftfahrzeugelektronik:
Grundlagen, Komponenten, Systeme, Anwendungen, Springer,
ISBN: 978-3-8348-9121-1, 2006
[Salmasi07] F. R. Salmasi, "Control strategies for hybrid electric vehicles:
Evolution, classification, comparison, and future trends" Vehicular
Technology, IEEE Transactions on 56.5 (2007): 2393-2404.

15

S-ar putea să vă placă și