phev control strategy optimization using matlab distributed … · phev control strategy...
TRANSCRIPT
PHEV Control Strategy Optimization Using MATLAB
Distributed Computing:From Pattern to Tuning
S. Pagerit, D. Karbowski, S. Bittner, A. Rousseau, P. SharerArgonne National LaboratorySponsored by Lee Slezak, U.S. DOE
G. SharmaThe MathWorks
MathWorks Automotive Conference3 June, 2008
2
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
3
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
4
New Constraints = More Complex Vehicle
Higher use of math-based tools before and during design:– Model-Based design– Physical modeling– Monte Carlo analysis– …
Caveat:– Increased detail of modeling, complexity– Increased number of simulations
=> Longer calculation, analysis and development time
EnvironmentFuel Economy HEVs and
PHEVsCost and
Complexity
5
More Complex Vehicle = More Sensitive Control
Common Strategy= EV+CS Blended Strategy
Optimization Evaluates Control Strategy’s Potential
s
EV MODE
lowICE
η
highICE
η
offoff
km/h
CS MODE
Veh. speed
SOC
%
Depending on various driven distance, several modes are possible during charge depleting: Electric-only (EV) and Blended
Higher Electric PowerHigher Electric Energy Higher Control
FreedomFuel Savings
Potential
s
%
BLENDED
km/h
BLENDED
highICE
η
off
highICE
η
off off
6
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
7
Tool: PSAT – Powertrain Systems Analysis ToolkitSetup
Simulation
Analysis
8
Process: 3-Way Approach to Control Optimization
( ) ( ){ }tSOCtP nm ,
( ) ( ){ } ( )tJtSOCtP '0,000 ,1',1 ++
( ) ( ){ } ( )tJtSOCtP MMMM ',' ,1,1 ++
( )tU 0
( )tU M
( )TX( )0X.
.
.
.
.
.
Optimal Control Pattern, Minimal
Fuel Cons.
Teng
Backward model Bellman Principle
Tmc
ωeng
ωmc
Optimally tunedParameters
Control Logic DIRECT Algorithm
PSAT Various PSAT Control Strategies
PSAT
Various Control
Principles Rule-Based Control Design
Global Optimization
Control Design
Derivative Free Optimization
9
Hardware: Computation Time Requires Distributed Computing
MATLAB and Parallel Computing Toolbox
MATLAB Distributed Computing Server - 1 Header Node - 2 Worker Nodes
MATLAB Distributed Computing Server - 2 Worker Nodes
…
# 1
# 2 # 8
Server Rack: - PSAT Software - Simulation Results
10
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
11
Finding Control Patterns Fast(er)
Robustness:– Different Cycles (e.g.: Urban, Highway,…)– Different Distances– Different Initial SOC
=> Set of 45 simulations=> Sequential computation time ~ 2 weeks
Using Distributed Computing:– Simulations run in parallel
=> Running time ~ 12 hours
12
Global Optimization Showed Minimal Fuel Consumption Achieved in Blended Mode
0 10 20 30 40 50 60 70 80 90 1000.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
% of total distance
SO
C
NEDC x10(gl. optim)UDDS x10 (gl. optim)Japan 10-15 x25 (gl. optim)Japan 10-15 x25 (PSAT)
Example of EV+CS Mode for comparison
3 cycles demonstrated blended strategy is optimal when knowing the distance
13
Global Optimization Showed Engine Starting Condition Almost Proportional to Electrical Consumption
0 50 100 150 200 2500
5
10
15
20
25
30
35
Electric Energy Consumption (Wh/km)
Pwr w
heel
s (kW
) abo
ve w
hich
P(IC
E=O
N)=
.95
Cycle:
HWFET
10 miles
20 miles 40 miles
Driven Distance:
UDDS
LA92
14
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
15
Engine ON/OFF Logic & Engine Torque Request
engine torque2
eng _pwr_dmd1
divideress_soc_init
nit .soc_
ess pwr charging
Operating regionfor the engine
f(u)
Eng hot opt torq curve
? ? ?
Eng hot low torq curve
? ? ?
Eng hot high torq curve
? ? ?
acc pwr6
eng trq max5
eng spd4
SOC 3
wh spd dmd2
wh trq dmd1
eng pwr dmd
eng_trq_opt_dmd
Engine ON / OFF Logic in Stateflow
Engine Torque Demand in Simulink
16
Blended Control Strategy Design Showed Significant Improvements Over EV Mode
-10%
-8%
-6%
-4%
-2%
0%
2%
0 16 32 48 64 80 96 112
Full Engine PowerOptimal Engine Power
Trip distance (km)
Up to 9% less fuel consumed
Fuel
Ene
rgy
Incr
ease
Ove
r EV/
CS
10 miles AER vehicle run on several UDDS cycles
17
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
18
Control Parameter Tuning with DIRECT
Robustness:– Different Cycles– Different Distances
Iterative Process:– Control space is sampled at each iteration– One simulation is run for each sample– Control space is re-sampled around the ‘best’ Simulation
=> Convergence after 30 iterations ~ 400 simulations=> Sequential computation time ~ 2 days
Using Distributed Computing:– Simulations run in parallel
=> Running time ~ 5 hours
19
The Longer the Electrical Distance, the More Robust
Influence of cycles (distance=23.7km)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
UDDS2 UDDS4 UDDS6 UDDS8
Fuel
Con
sum
ptio
n R
atio
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Bat
tery
Ene
rgy
Rat
io
Fuel Consumption RatioBattery Energy Ratio
Influence of cycles (distance =71.2Km)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
UDDS2 UDDS4 UDDS6 UDDS8Cycles
Fuel
Con
sum
ptio
n R
atio
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Bat
tery
Ene
rgy
Rat
io
Fuel Consumption RatioBattery Energy Ratio
Tuned for a short distance (2xUDDS)
Tuned for a longer distance (6xUDDS)
20
Outline
IntroductionSetupGlobal Optimization for PatternsReal Time ControllerDIRECT Optimization for TuningConclusion
21
Conclusion
Using a combination of optimization techniques and modeling based on PSAT and MATLAB, we were able to:– single out control patterns,– implement them in Simulink and Stateflow,– tune their parameters.
Only the use of distributed computing allows this process to be performed in a timely manner:– After setting up the MATLAB Parallel Computing Toolbox, and
Distributed Computing Servers, it took only 1 hour of development to get the first simulations running.
– The optimization times were reduced from more than 2 weeks to less than a day.
22
ReferencesPlug-in Hybrid Electric Vehicle Control Strategy: Comparison between EV and Charge-Depleting OptionsSharer, P. / Rousseau, A. / Karbowski, D. / Pagerit, S.SAE World Congress 2008, April 2008
Impact of Component Size on Plug-In Hybrid Vehicle Energy Consumption Using Global OptimizationKarbowski, D. / Haliburton, C. / Rousseau, A. EVS 23, December 2007
Plug-in Hybrid Electric Vehicle Control Strategy Parameter Optimization Rousseau, A. / Pagerit, S. / Gao, D. EVS 23, December 2007
Plug-in Vehicle Control Strategy: From Global Optimization to Real Time ApplicationKarbowski, D. / Rousseau, A. / Pagerit, S. / Sharer, P. EVS22, October 2006.
Global Optimization to Real Time Control of HEV Power Flow: Example of a Fuel Cell Hybrid VehiclePagerit, S. / Rousseau, A. / Sharer, P.EVS 21, April 2005.