Dept. Of Mechanical and Industrial Engineering University of Illinois at UrbanaChampaign

1
Modeling and Multivariable Control of An
Earthmoving Vehicle PowertrainPart I Modeling

• Prof. Andrew Alleyne
• Rong Zhang
• Eko Prasetiawan
• Caterpillar Inc.
• June 13th, 2001
• Technical Center, Caterpillar Inc., Peoria

advisor speaker ex-team partner project
sponsor
2
Part I Modeling

• 1. Introduction
• Background and mission
• 2. Method
• 3. Results and Discussion
• Powertrain model 14 states, I/O5/9
• Powertrain simulator
• 4. Conclusions

3
1. Introduction
Introduction Method Results
Discussions Conclusions

• Background
• The earthmoving vehicle powertrain
• Potential benefits of powertrain control
• The earthmoving vehicle powertrain simulator A
  Hardware-In-The-Loop facility
• Mission of modeling

4
Earthmoving Vehicle Powertrain
Introduction Method Results
Discussions Conclusions
5
Benefits of Powertrain Control
Introduction Method Results
Discussions Conclusions

• Quality Performance
• Control coordination and ease of use
• Cost efficiency
• Taking advantage of electro-hydraulic technology
• Design simplification
• Energy efficiency
• Power optimization

6
EVPS testbed
Introduction Method Results
Discussions Conclusions
7
EVPS Schematic
Introduction Method Results
Discussions Conclusions
8
Mission of Modeling
Introduction Method Results
Discussions Conclusions

• Powertrain Control
• Model-based controller design
• Simplified system model
• Simulator Control
• Component emulator design
• More detailed reference models and actuator models

9
2. Method
Introduction Method Results
Discussions Conclusions

• How to model the powertrain?
• Subsystem analysis modeling of components
• System synthesis Integration of subsystem models
  in the State Space
• How to simulate the powertrain in experiments?
• Engine emulation speed tracking
• Load emulation Model Reference Control

10
Subsystem analysis
Introduction Method Results
Discussions Conclusions

• Qualitative dynamics
• First principles Published literature
• Quantitative parameters
• Experimental identification Scaling estimation
• Components
• Engine, Pump, Flow Valve, Lumped Volumes, Motor,
  Load
• Pressure valve

11
Engine Model
Introduction Method Results
Discussions Conclusions

• The engine
• For the Virtual Engine
• 3-states, nonlinear, SI engine Moskwa and
  Hedrick, 1987
• For linear controller design
• 2-states, linear model
• Model inaccuracy taken as uncertainty of the
  linear controller
• A CI engine model will be used later.

Engine Inertia
Throttle ? Torque
Viscous Damping
12
Pump Model
Introduction Method Results
Discussions Conclusions

• The pump

Swashplate angle ref.
Swashplate dynamic
Flow output
Variable displacement

• The hose (assumed rigid)
• A lumped volume

Bulk modulus
Pressure rise-rate
Net input flow
Hose volume
13
Flow Valve the Valvistor
Introduction Method Results
Discussions Conclusions
14
Flow Valve Model
Introduction Method Results
Discussions Conclusions

• The flow valve
• 2 Mass-spring-damping
• 4th order, 2 zeros
• Reduced as 1st order, 1 zero
• Electronic dynamics ignored

15
Hydraulic Motor Model
Introduction Method Results
Discussions Conclusions

• Free of external load a faster 1st order (for
  control design)
• General load nonlinear dynamics depending on
• Numerical example

Viscous damping
Motor inertia
Load torque
16
Load Models
Introduction Method Results
Discussions Conclusions

• Traction load
• Motor Vehicle Tire Model ( Working cycle)
• (Power steering load)
• (Implement load)

17
Traction Load Model
Introduction Method Results
Discussions Conclusions
18
Traction Load
Introduction Method Results
Discussions Conclusions
For a scaled vehicle equivalent to the EVPS
19
Load Engine Emulation
Introduction Method Results
Discussions Conclusions

• Model Reference Control

20
Pressure Valve model...
Introduction Method Results
Discussions Conclusions
Mainvalve
Pilot Valve
Speedsensor
Pressuresensor
21
...Pressure Valve Model
Introduction Method Results
Discussions Conclusions

• 2 mass-spring-damper systems
• 4th order dynamics
• Simplified as a 2nd order system
• Nonlinearity DC gain varies with input voltage

22
...Load Emulation...
Introduction Method Results
Discussions Conclusions

• Simulation of the direct application of MRC

Original Plant
Controlled Plant
Reference Model
23
...Load Emulation...
Introduction Method Results
Discussions Conclusions

• With an adaptive online simulation

Controlled Plant
-
-
Compres.
Motor Speed
Gear Motor
Controlled Valve
Dm
Reference
24
Engine Emulation
Introduction Method Results
Discussions Conclusions
Torque (N-m)
PCEngineEmulator
AC Motor
ABBController
EngineModel
Virtual Throttle (deg)
Speed (rpm)
Speed Ref. (rpm)
Speed (rpm)
A fast plant
A slow reference model
25
3. Results and Discussion
Introduction Method Results
Discussions Conclusions

• System modeling
• State space representation
• Experimental verification
• Load emulation
• Performance in tracking reference model

26
System synthesis
Introduction Method Results
Discussions Conclusions
27
State-space representation
Introduction Method Results
Discussions Conclusions
28
Model verification
Introduction Method Results
Discussions Conclusions
UpstreamPressure
Swashpl.angle
(Step input)
Experiment
Simulation
LoadSpeed
Downstreampressure
29
Load Emulation
Introduction Method Results
Discussions Conclusions

• Experimental results
• Without adaptation
• With adaptation

30
4. Conclusion
Introduction Method Results
Discussions Conclusions

• A powertrain model
• A linear system at an operating point
• Ready for controller design
• A powertrain simulator
• Dynamics of the powertrain emulated by a testbed
• Ready for controller evaluation
• Future topics
• System ID at other operating points
• Engine emulation Load emulation for other
  dynamics