model -based powertrain and engine control · model -based powertrain and engine control guoming...
Post on 20-Mar-2020
15 Views
Preview:
TRANSCRIPT
ModelModel--Based Powertrain and Engine ControlBased Powertrain and Engine ControlGuoming (George) ZhuGuoming (George) Zhu
Mechanical Engineering, Electrical and Computer EngineeringMechanical Engineering, Electrical and Computer Engineering
Michigan State UniversityMichigan State University
(7(7--44--2013)2013)
07/04/2013 Page 2Zhu
� Background
� Model-based control frame work
� Modeling
� Crank-resolved engine/powertrain model
� Control oriented charge-mixing model
� Control Applications
� Optimal control: HCCI model transition control
� LPV (Gain-scheduling) control of cam phaser
� Adaptive control of LNT regeneration
� Conclusions
OutlineOutline
07/04/2013 Page 3Zhu
BackgroundBackground
VGT/Super charger Cam phasing
(camless)
Wastegate
ETC
PlantPID
Traditional SISO Control
MIMO Control System
PlantMBC
Engine system has the following characteristics:
� Highly nonlinear
� Parameter (time) varying
� System parameters changes as engine aging
� Time and event based control
� Complicated flow and thermal dynamics
� Robustness (fuel, altitude, etc.)
Model-Based Control (MBC) is
a necessity
PFI/DI
ignition
Oxygen/Nox sensors
Pressure/temperature
sensors
Cylinder pressure
and combustionsensors
07/04/2013 Page 4Zhu
ModelModel--Based Control FrameworkBased Control Framework
Dynamic
System
Disturbance
System Outputs
Disturbance
Estimation
and/or
Model Adaptation
System
Model
Model-Based
Controller or
Optimizer
Reference Signal
Updated Model and Disturbance
MeasurementsControl
Generic form of the model-based control
Two key components of model-based control:
�control oriented model capable of real-time simulation, and
�model based controller and optimizer
07/04/2013 Page 5Zhu
ModelModel--Based Control Based Control (development roadmap)(development roadmap)
Model-based
control strategy
Crank-Based
Combustion
Model�EGR
Πlift
θINTM
θEXTM
θST
Scheduled
Open-loop
Control
PI Control
(Feedback)
ILC Control
LQ Control
Memory
+
+
IETC
FDIλ(i+1) IMEP(i+1)
IMEP(i)
λ(i)
FFF(i)
FILC
FFB
MAP
�TPS
Time-Based
Air Handling
Model
FFF(i+1)
HIL Engine
SimulatorPrototype Engine Controller
IMEP
Control-oriented engine models
developed with GT-Power
simulations
HIL simulation to
validate the
control strategy
Single cylinder optical engine
Multi-cylinder metal engine
dSPACE based
real-time
simulator
Harness
break-out box
Opal-RT
based engine
prototype
controller
dSPACE host computer
Opal-RT host computer
Simulated signal display oscilloscope
����
07/04/2013 Page 6Zhu
ControlControl--Oriented ModelingOriented Modeling
Dynamic System
Fundamental(first principle
model)
Data-Driven(empirical model)
Nonlinear Diff. Equations.
(ODEs & PDEs)
Reduced Order Models
Discrete-Time State Space
Models
Continuous-Time State
Space Models
Frequency Domain
Transfer Funs
Linearized Models
Linearized Empirical Models
(e.g., ANN)
Linearization
07/04/2013 Page 7Zhu
CrankCrank--Resolved Engine ModelResolved Engine Model
� Crank-based in-cylinder pressure and temperature;
� Simple in-cylinder charge mixing model;
� Event-based fuel, ignition, and torque calculation with mean manifold dynamics.
� Real-time simulation up to 5000RPM
07/04/2013 Page 8Zhu
PhysicsPhysics--based modeling: based modeling: charge mixing (1)charge mixing (1)
BackgroundBackground
140° after exhaust TDC 180° after exhaust TDC 120° before comb TDC 60° before comb TDC
� HCCI combustion assumes Homogeneous Charge before compression
ignition, while in practice it is Heterogeneous, especially with high EGR.
� One-zone control-oriented HCCI combustion model, developed earlier,
assumes that the thermodynamic characteristics is uniformly distributed in the
cylinder, leading large prediction error of the start of combustion (SOC).
� To accurately predict the SOC, it is proposed to use a two-zone HCCI
combustion model for predicting SOC, which involves two-zone charge mixing
and HCCI modeling.
* M. Shen, “Simulation of in-cylinder flow and composition distribution of a gasoline HCCI engine with variable valve actuation,” MS Thesis, Tianjin University, July, 2006.
07/04/2013 Page 9Zhu
Charge mixing modeling - three phases:
� Backflow phase: when in-cylinder pressure
is higher than manifold pressure;
� Backflow returning: when in-cylinder
pressure is lower than manifold pressure;
� Fresh charge phase.
Diffusion
• Molecular diffusion
• Laminar diffusion
• Turbulent
diffusion
Mass transfer between fresh charge
and residual is assumed to be mainly
due to
350 400 450 500 550 600 650 700-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Crank position (deg)
Mass
(g)
Mixed-zone mass
Unmixed-zone mass
Total trapped mass
SOCIVCIVO
stage 1 stage 2 stage 3
PhysicsPhysics--based modeling: based modeling: charge mixing (2)charge mixing (2)
* S. Zhang, G. Zhu, and Z. Sun, “A control-oriented charge mixing and two-zone HCCI combustion model," Submitted to IEEE Transactions on Vehicular Technology (Feb., 2013).
07/04/2013 Page 10Zhu
Model
FUE
L
(MG)
SOC IMEP
GT-Power 13.2 2 4.2
Two-zone model 13.2 2 4.22
One-zone model (w/ flow dynamics) 13.2 4 4.23
One-zone model (w/o flow dynamics) 13.2 8 4.36
Pre
ssu
reT
em
pera
ture
-100 0 100 200 300 400 5000
5
10
15
20
25
30
35
40
Crank position (deg)
In-c
ylin
der
pre
ssure
(bar)
GT-Power modelTwo-zone model
-100 0 100 200 300 400 5000
500
1000
1500
2000
Crank position (deg)
In-c
ylin
der
tem
pera
ture
(K
)
GT-Power modelTwo-zone model
200 250 300 350 400 450 5000
2
4
6
8
10
Crank position (deg)
In-c
ylin
der
pre
ssure
(bar) GT-Power model
Two-zone model
One-zone model
200 250 300 350 400 450 500
400
600
800
1000
1200
1400
Crank position (deg)
In-c
ylin
der
tem
pera
ture
(K
) GT-Power model
Two-zone model
One-zone model
In-Cylinder Mass Flow Rate (GT-Power and Two-Zone Models)
100 150 200 250 300 350 400 450 500-0.05
0
0.05
0.1
-0.05
0
0.05
0.1
-0.05
0
Crank position (deg)
Ma
ss fl
ow
rate
(kg/s
)
GT-Power model
Two-zone model
PhysicsPhysics--based modeling: based modeling: charge mixing (3)charge mixing (3)
07/04/2013 Page 11Zhu
Hybrid powertrain Hybrid powertrain –– HIL simulationsHIL simulations
Hybrid powertrain supervisory
control for the best fuel economy
using real-time equivalent fuel
consumption optimization
Real-time hybrid powertrain modeling for the
hardware-in-the-loop (HIL) simulation applications
CAN
Mototron based
supervisory controller
Opal-RT based
real-time simulator
TruckSim + Driver Model
gal
total fuel used (gal)
drive cycle
Vehicle
Speed
Power bus MPG
Gal
Engine_on
key_on
time
Fuel converter
Gal
Distance
Energy storage
EMB
EMA
Driver
Divide
Controller
1
Constant1
1
Constant
ClockAHS1
TruckSim + Driver Model
07/04/2013 Page 12Zhu
Optimal Control: Optimal Control: HCCI Mode TransitionHCCI Mode Transition
Two cycle throttle pre-opening
to prepare for valve lift switch
from high to low
LQ Optimal throttle control
to maintain desired AFR
Valve lift switch
from high to low
Hybrid combustion to match
increased recompression due
to electric cam phasing
Hybrid combustion that starts with SI
and ends with HCCI combustion
DI fuel control to ensure smooth mode transition
-80 -60 -40 -20 0 20 40 60 80
0
0.2
0.4
0.6
0.8
1
Crank angle (deg)
Mass Fraction Burned
ST
SOHCCI
-80 -60 -40 -20 0 20 40 60 80
0
0.2
0.4
0.6
0.8
1
Crank angle (deg)
Mass Fraction Burned
ST
SOHCCI
-80 -60 -40 -20 0 20 40 60 80
0
0.2
0.4
0.6
0.8
1
Crank angle (deg)
Mass Fraction Burned
SOHCCI
* X. Yang and G. Zhu, “SI and HCCI combustion mode transition control of a multi-cylinder HCCI capable SI engine,” IEEE Transaction on Control System Technology (Accepted in May, 2012, DOI: 10.1109/TCST.2012.2201719)
07/04/2013 Page 13Zhu
LPV (GainLPV (Gain--Scheduling) Control Scheduling) Control
� Traditionally, the PID control gains are tuned by calibration engineers in test cell
or field.
� Control design based upon a linear system model whose parameters are a
function of measurable parameters; and the resulting controller parameters are
also a function of these measurable parameters.
� Closed loop system stability and performance are guaranteed
* A. White, Z. Ren, G. Zhu, and J. Choi, “Mixed H∞
and H2 LPV control of an IC engine hydraulic cam phase system,” IEEE Transaction on Control System Technology, Vol. 21, Issue. 1, 2013, pp. 229-238 (DOI 10.1109/TCST.2011.2177464)..
VVT example VVT example ( 1) ( ) ( ) ( ) ( )
( ) ( ) ( )
c c c
c
x k A x k L z k
u k K x k
θ θ
θ
+ = +
=
07/04/2013 Page 14Zhu
Adaptive Control: Adaptive Control: AFR during LNT regenerationAFR during LNT regeneration
� Biofuel content is estimated online with guaranteed convergence
� Optimal air-to-fuel ratio (AFR) tracking control as a function of biofuel content
� Guaranteed closed loop system stability under any fuel content
• X. Chen, Y. Wang, I. Haskara, and G. Zhu, “Optimal air-to-fuel ratio tracking control with adaptive biofuel content estimation for the LNT regeneration,” IEEE Transaction on Control System Technology (Accepted in March, 2013, DOI: 10.1109/TCST.2013.2252350).
EngineExhaust
Manifold
Oxygen
Sensor
fuelm Φˆ
ˆair
DS
m
ασ
DS
airm
ασ
Plant
3x2
x1xu
System Model0.6
0.8
1
1.2
Fu
el G
ain
100 200 300 400 500 600 700 800
1
1.2
1.4
λ
350 400 450 500
1
1.2
1.4
Time(s)
λ
Fuel Gain
Reference λ
Measured λ
Reference λ
Measured λ
Plant
Adaptive Fuel
Gain Estimation
LQ
Contoller
State
Estimation
r
x̂
3x
α̂
Estimate
Fuel Mass
fuelmu
07/04/2013 Page 15Zhu
� Model-based powertrain/engine control becomes a necessity due to the
significant increment of number of sensors/actuator and high system
nonlinearity.
� Control-oriented powertrain and engine modeling is moving towards
first-principle based with reduced complexity (e.g., engine charge-
mixing model)
� Powertrain and engine models used for the HIL (hardware-in-the-loop)
simulations will be capable of simulating the physical systems at
different detail level. The improved computing technology enables more
and more first-principle based simulations.
� Model-based control, such as adaptive control, model predictive control,
linear parameter varying (gain-scheduling) control, will be the future
powertrain and control technologies
ConclusionsConclusions
07/04/2013 Page 16Zhu
� Closed loop combustion control of internal combustion engines (SI,
HCCI, and CI)
� Adaptive and model reference control of hydraulic and electric valve
actuation
� Closed loop system identification and control of automotive systems
� Hybrid powertrain system control and optimization
� Automotive system modeling for hardware-in-the-loop (HIL) simulations
� Combustion control and optimization for ethanol engines
� Variable displacement engines
� Ionization based combustion diagnostics and control
� TEG (thermo-electric generator) system management
� Application of the smart material to automotive systems
� LPV (linear parameter varying) optimal control with hard constraints
Other Research ActivitiesOther Research Activities
top related