Designed by Deokjin Kim ( [email protected] )
2019AVL
InternationalSimulation Conference
AVL
Crui
seOptimal supervisory control strategyfor a transmission-mounted electricdrive (TMED) hybrid electric vehicle
Ph.D. Taeho ParkSenior Researcher
Advanced Powetrain R&D CenterKorea Automotive Technology Institute (KATECH)
2019. 10. 22
Designed by Deokjin Kim ( [email protected] )
INDEX
1. Introduction
2. System modeling
3. Optimal supervisory control based on ECMS
4. Simulation results
5. Conclusion
Designed by Deokjin Kim ( [email protected] )
1. Introduction1.1 Fuel economy improvement of the full-type HEV
• Regeneration braking : reduces the friction brake loss• Idle stop & go : reduces the fuel consumption during the engine idling• Engine operating point control (Load leveling, ECMS, etc.) & EV mode
: moves engine operating points to the high efficiency region (near OOP or OOL): removes low efficiency operation of the engine by EV mode
Fuel-tank Hybrid Propulsion System
Differential Case
Wheel
WheelEnergy Storage
Device
Engine Poweron the OptimalOperation Point
Electric Power : Difference betweenEngine Power and Wheel Power
Wheel Power
Engine operating point control
Idle Stop & Go
Combustion Engine
EV mode
Brake
BrakeRegenerative
Braking Energy
OptimalOperating
Point (OOP)
OptimalOperating
Point (OOP)
Designed by Deokjin Kim ( [email protected] )
1. Introduction1.2 Representative configuration of parallel HEV
FMED (P0, P1) TMED (P0+P2) P0 + P3
Structure
Idle stop & go O O O
EV mode X O (Limited if T/C is applied) O
Engine load control O O O
Regeneration braking O (Limited) O O
Series hybrid mode X O O
Required no.of MGs 1 2 2
è Target vehicle configurationè Target vehicle configuration
• Target vehicle configuration : Transmission mounted electric drive (TMED, P0 + P2) HEV
Designed by Deokjin Kim ( [email protected] )
1. Introduction1.3 Classification of the supervisory control algorithm
F. R. Salmasi, “Control Strategies for Hybrid Electric Vehicles Evolution, Classification, Comparison, and Future Trends,” IEEE TVT, 2007
• Target algorithm : Equivalent (Fuel) Consumption Minimization Strategy (ECMS)
Designed by Deokjin Kim ( [email protected] )
1. Introduction1.4 Introduction to equivalent consumption minimization strategy (ECMS)
• Equivalent consumption: Actual fuel consumption + estimated fuel consumption for charge compensation in future
• Equivalence factor: ratio between the fuel power for charge compensation and present electric power consumption
Discharge CaseCharge Case
Instantaneous cost function :
Equivalence factor
G. Paganelli, G. M. Guerra, S. Delprat, J-J Santin, M. Delhom, and E. Combes,“Simulation and assessment of power control strategies for a parallel hybrid car,” IMechE part D, 2000
Designed by Deokjin Kim ( [email protected] )
1. Introduction1.5 Mode transition operation of TMED HEV
• Problem of the quasi-stationary model based numerical optimization : Frequent mode transition can occur for optimizing steady-state efficiency optimization
• Mode transition operation of TMED HEV: Step 1 – Engine cranking: Step 2 – Clutch input / output speed synchronization: Step 3 – Clutch engagement: Step 4 – Torque command transition
0 100 200 300 400 500 600 700 800 900 1000 11000
50
100
Veh S
peed [
kph]
0 100 200 300 400 500 600 700 800 900 1000 11000
0.5
1
Fla
g [
-]
BoostClutchRelease
0 100 200 300 400 500 600 700 800 900 1000 11000
2000
Speed [
rpm
]
EngineMG2
0 100 200 300 400 500 600 700 800 900 1000 1100-500
0
500
time [s]
Torq
ue [
Nm
]
EngineMG2MG1
0 100 200 300 400 500 600 700 800 900 1000 110020
40
60
time [s]
SO
C [
%]
0 100 200 300 400 500 600 700 800 900 1000 11000
50
100
Veh S
peed [
kph]
0 100 200 300 400 500 600 700 800 900 1000 11000
0.5
1
Fla
g [
-]
BoostClutchRelease
0 100 200 300 400 500 600 700 800 900 1000 11000
2000
Speed [
rpm
]
EngineMG2
0 100 200 300 400 500 600 700 800 900 1000 1100-500
0
500
time [s]
Torq
ue [
Nm
]
EngineMG2MG1
0 100 200 300 400 500 600 700 800 900 1000 110020
40
60
time [s]
SO
C [
%]
è Excessive energy consumption can occur in the commercial vehicle which
has large rotational inertia of the engine
è Excessive energy consumption can occur in the commercial vehicle which
has large rotational inertia of the engine
Designed by Deokjin Kim ( [email protected] )
1. Introduction1.6 Limitation of previous researches / Contribution of the proposed method
• Limitation of previous researches on ECMS: Some mode transition operations such as clutch speed synchronization is not considered
: Penalty term for mode determination algorithm requires tuning parameters which need to be optimized for the driving pattern
: The instantaneous cost function which is calculated at each 1 ~ 10ms interval is directly compared with the penalty term of mode transition which lasts few hundreds of milliseconds
• Contribution of the proposed ECMS: Adopts the engine acceleration model to exactly calculate the fuel energy and the electrical energy consumed during the engine cranking and synchronization
: Cost functions for all driving modes of TMED HEV are incorporated in the ECMS
: Calculation intervals of the mode transition energy and the instantaneous cost function are synchronized by adopting integral type cost function
Designed by Deokjin Kim ( [email protected] )
2. System modeling2.1 Forward simulation model of TMED HEV (AVL Cruise®)
• Hybrid propulsion device : engine, MG1(HSG), MG2(Traction motor)• Electric energy storage : HV battery• Driveline dynamics : clutch, automated manual transmission (AMT), differential gear• Subsystem controller : AMT controller with GB control / program, Anti-slip control
MG1
MG2
Diffe
rential
Gea
r
Engine
Engine Clutch
Inverter
Battery
AMTGearbox
EV path
Series path
Parallel path
Designed by Deokjin Kim ( [email protected] )
2. System modeling2.2 Quasi-stationary model of TMED HEV
• Dynamic model : Battery SOC dynamics• Stationary model : Performance curve of engine, MG1, MG2, and battery
: Control input 1 (u1) – Mechanical power split ratio
: Control input 2 (u2) – Electrical power split ratio
PBat.loss MG2
FuelTank
Battery MG1
Engine
Gearbox& Axle
PMG2
PMG1
HLHV mf
TMG2·ωMG2
Teng·ωeng
TMG1·ωMG1 ClutchPBat
u1
u2
des
MG
engMG
MG
engengMGMG
MGMG
des
MGMG
TT
TTT
TTT
PTu 2
2
2
22
22221 =
+=
+=º
wwww
2
21
22
MG
MGMG
MG
bat
PPP
PPu +
=º
u1 u2 Driving mode
EV mode
Series hybrid mode
Parallel hybrid mode
- Not used -
11 =u12 ¹u11 =u12 =u11 ¹u12 ¹u11 ¹u
12 =u
Driving mode according to power split ratios
Designed by Deokjin Kim ( [email protected] )
• Engine friction torque
• Engine rotational dynamics
• Engine speed
• Fuel power
• Electric power
2. System modeling2.3 Engine acceleration energy model (*Engine clutch is disengaged)
cengvengfric ftfT += )()( ww
)()()()( 11
2engfricengMGengeng TtTtT
NNtJ ww -+=&
ò÷÷÷
ø
ö
ççç
è
æ
-+=
t
accengfricaccengeng
accMGMG
engacceng dt
TT
TNN
Jt
0
)()(
)(1)(
..max.
.1max.12
1
.
ww
ww
)),(()( ..max.. accengaccengengFCLHVaccf TfHtP ww=
)()()( ..1. tPtPtP acclossbatMGacce += ( )busbus
MG
busbus
MGMG
busbus
MG
busbusacclossbat
iSOCrtVtP
iSOCrtVtPtP
iSOCrtVtP
iSOCrtitP
,)()(
),()()()(2
),()()(
),()()(
int
2
1
int221
int
2
2
int2
..
÷÷ø
öççè
æ+
÷÷ø
öççè
æ=
÷÷ø
öççè
æ-
º
TMG1, ωMG1
Teng, ωeng
N1
N2 Jeng
Tfric
( )( ) ( ) cengvengfric ftftT += ww
• Battery power loss
Designed by Deokjin Kim ( [email protected] )
3. Optimal supervisory control based on ECMS3.1 Structure of proposed algorithm
• Driving status determination (DSD): generates internal variables and flags
• Time windowed equivalent consumption minimization strategy (ECMS): determines optimal power split ratios and mode flags using ECMS
• Powertrain status management (PSM): determines present driving mode and final output commands
Driving Status Determination
(DSD)Vehicle State
Flags
Tdes
Optimal Power Split-ratio (u1.opt, u2.opt)
DrivingModeVehicle
Signal
Lever Signal
Pedal Signal
Engine Command
Motor Command
Transmission Command
OptimalMode Flags
Powertrain StatusManagement (PSM)- Mode determination- Power distribution
Optimal Driving Mode Determination Algorithm
Optimal Power Distribution Algorithm
Optimal cost functions & MG1 Speed command
(ΔJEV, ΔJSeries.opt, ΔJParallel.opt, ωMG1.Series.opt)
Time windowed Equivalent Consumption Minimization Strategy (ECMS)
Designed by Deokjin Kim ( [email protected] )
3. Optimal supervisory control based on ECMS3.1 Structure of proposed algorithm
• Types of ECMS simulated in this research
Abbreviationin this
research
Powerdistribution algorithm
DrivingMode Mode determination algorithm
OriginalECMS Original
Original ECMS
EV modeParallel mode
Instantaneous cost comparison
Previous algorithm type A Prev-A Instantaneous cost comparison with
constant penalty for mode transition
Previous algorithm type B Prev-B Instantaneous cost comparison with
proportional penalty for mode transition
Proposed algorithm type A Prop-A
EV modeParallel modeSeries mode
Integral type cost comparison with detailed engine acceleration energy calculation(Time window length = Mode transition time)
Proposed algorithm type B Prop-B
Integral type cost comparison with detailed engine acceleration energy calculation(Time window length = Fixed tuning factor)
Designed by Deokjin Kim ( [email protected] )
4. Simulation results4.1 Simulation environment
• Simulation environment : AVL Cruise® / Simulink® co-simulation• Driving mode for performance evaluation : City (FTP72) / Combined (WHVC)
Vehicle Model (Cruise®)
EMSCommand
MG / ISG / LDCCommand
TCUCommand
EMS Signal
TCU Signal
MCU / GCU / LDC/ BMS Signal
HCU algorithm (Simulink®)
0 200 400 600 800 1000 1200 1400 16000
20
40
60
80
100
Veh
icle
spe
ed (k
m/h
)
Time (s)
0 200 400 600 800 1000 12000
20
40
60
80
100
Veh
icle
spe
ed (k
m/h
)
Time (s)
Urban: 5.3km
Rural: 5.8km
Motorway:8.9km
Designed by Deokjin Kim ( [email protected] )
4. Simulation results4.2 Control behavior of proposed ECMS (City cycle)
• Section A: EV to Parallel mode change due to a high driver demand torque (Boost flag)
• Section B: EV to Parallel mode change due to the optimal driving mode by proposed ECMS
• Section C: Parallel to EV mode change considering both conditions: Even though the driver demand torque can be driven only by MG2, the mode is kept in the
parallel mode which is the optimal driving mode by proposed ECMS (1273s ~ 1276.5)
18 20 22 24 26 28 300
50
Veh
Spe
ed [k
ph]
18 20 22 24 26 28 300
0.5
1
Flag
BoostClutch
18 20 22 24 26 28 300
2000
Spe
ed [r
pm]
EngineMG2
18 20 22 24 26 28 30-500
0
500
time [s]
Torq
ue [N
m]
EngineMG2MG1
18 20 22 24 26 28 300
1
2x 10
6
Cos
t [J]
EVSeriesParallel
998 1000 1002 1004 1006 1008 101035
40
45V
eh S
peed
[kph
]
998 1000 1002 1004 1006 1008 10100
0.5
1
Flag
BoostClutch
998 1000 1002 1004 1006 1008 10100
1000
2000
Spe
ed [r
pm]
EngineMG2
998 1000 1002 1004 1006 1008 1010-500
0
500
time [s]
Torq
ue [N
m]
EngineMG2MG1
998 1000 1002 1004 1006 1008 1010
0246
x 105
Cos
t [J]
EVSeriesParallel
1270 1275 1280 128520
30
40
Veh
Spe
ed [k
ph]
1270 1275 1280 12850
0.5
1
Flag
BoostClutch
1270 1275 1280 12850
100020003000
Spe
ed [r
pm]
EngineMG2
1270 1275 1280 1285-200
0200400600800
time [s]
Torq
ue [N
m]
EngineMG2MG1
1270 1275 1280 12850
1
2
x 106
Cos
t [J]
EVSeriesParallel
Designed by Deokjin Kim ( [email protected] )
4. Simulation results4.3 Quantitative result comparison (City cycle)
• Fuel economy / Fuel use : Proposed algorithms show best fuel economy• Number of mode transitions : Proposed algorithms show optimized mode transition behavior• Share of cycle time : Proposed algorithms maximize the EV mode driving
Designed by Deokjin Kim ( [email protected] )
Cou
nt (-
)
4. Simulation results4.4 Engine operation point analysis (City cycle)
Cou
nt (-
)
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Engine efficiency (-)
0
500
1000
1500
2000
2500
Cou
nt (-
)
ParallelSeriesModeTransStanding
• Original ECMS : Some low-efficiency operating points exist• Prev-A & Prev-B : Low-efficiency operating points are increased• Prop-A & Prop-B : Low-efficiency operating points are reduced compared with original ECMS
LOW ç è HIGH
LOW ç è HIGH
LOW ç è HIGH
LOW ç è HIGH
LOW ç è HIGH
Designed by Deokjin Kim ( [email protected] )
5. Conclusion
1. An optimal control framework consideringentire power flow of TMED HEV is introduced
based on the time-windowed ECMS
2. Excessive mode transition is reducedby the proposed mode determination algorithm
based on the engine acceleration energy
3. Both types of proposed algorithm show best fuel economycompared to previous algorithms,
even though tuning parameters of the mode determination algorithmare reduced (Prop-B) or eliminated (Prop-A)
Designed by Deokjin Kim ( [email protected] )
2019AVL
InternationalSimulation Conference
AVL
Crui
seOptimal supervisory control strategyfor a transmission-mounted electricdrive (TMED) hybrid electric vehicle
2019. 10. 22
Thank you
Ph.D. Taeho ParkSenior Researcher
Advanced Powetrain R&D CenterKorea Automotive Technology Institute (KATECH)