|
1
|
- Chris Foster and J.Q. Sun
- Department of Mechanical Engineering
- University of Delaware
|
|
2
|
- Introduction
- Objectives
- Collaboration
- System Hardware
- Lower Level Hardware Controllers
- Electrical Current Control
- Cylinder Position Control
- Intelligent Velocity Controllers
- PID Velocity Control
- Supervisory Velocity Control
- Adpative Velocity Control
|
|
3
|
- What’s driving electronics in vehicles?
- Consumer expectations
- Automation
- Simplification of user interfaces
- Increase productivity
- X-by-Wire technologies
- X-by-Wire systems allow for electronic actuation of hardware.
- This enables the implementation of electronic controls to automate
systems.
|
|
4
|
- Fully Automated vehicles
- Vision systems for navigation and obstacle detection.
- A variety of sensors to provide information for decision making.
- Stratified software and control structure
- Higher level control systems (Task manager) can request adjustment=
of
specific vehicle parameters by lower level controllers.
- X-by-Wire solutions for controlling hardware components.
|
|
5
|
- Collaborative project between NASA, New Holland North America LLC. a=
nd
Carnegie Mellon University to build a fully autonomous agricultural
vehicle.
- Vehicle successfully cut 120 Acres of crop on its own.
|
|
6
|
- Velocity Control of SP Windrower
- Development of a cost effective velocity control solution for a
self-propelled (SP) windrower.
- Investigate intelligent control systems to automatically compensate=
for
hardware variations.
- Develop associated hardware to implement the control systems.
|
|
7
|
|
|
8
|
|
|
9
|
- Objective
- To regulate the ground speed of the vehicle as sensed by the wheel
speed sensors.
|
|
10
|
|
|
11
|
- Simulation Data – Low ratio
|
|
12
|
- Experimental Data – Low ratio
|
|
13
|
- PID control stabilized the system and showed robustness.
- If tuned for acceleration system exhibited excessive overshoot in
deceleration.
- If tuned for deceleration system exhibited a sluggish response in
acceleration.
- Can conclude that the plant dynamics are changing between accelerati=
on
and deceleration.
|
|
14
|
- Possible varying parameters
- Hydraulic efficiency/leakage of motors and pumps.
- Dynamics of relief valves.
- Changing oil viscosity with oil temperature.
- Variations in cylinder position control dynamics.
- Size of tyres on vehicle.
- Type of header on vehicle (width, disc/sickle/draper).
- Suspension configuration of vehicle.
|
|
15
|
- Multi-Estimator
- Contains a family of plant models.
- Generates set of predicted tracking errors (ep).
- Monitoring signal generator
- Provide a set of performance signals (mp) based on (ep),
to the supervisor.
- Supervisor
- Uses monitoring signals (mp) to decided which plant model
best matches response of plant at present time.
- Decides when to switch controllers.
- Multi-controller
- Contains a family of controllers that stabilize plant models in the
multi-controller.
|
|
16
|
|
|
17
|
- Selects which controller to implement based on information in monito=
ring
signals (mp)
- Scale independent switching logic
- Prevents fast switching
- Filters out short term transient effects
- Switching unchanged if mp scaled by a positive function.=
|
|
18
|
- Define scaled performance signal
- If we replace with  =
;
switching of s will be unchanged.
-  =
;
is a monotonically non-decreasing function.
- If  =
;
and  =
;
as  =
;
then switching will stop in finite time.
|
|
19
|
|
|
20
|
|
|
21
|
- Performed well overall
- Removal of short term switching of s will improve performance.
- Step tests induce undesirable dynamics.
- Engine RPM can be incorporated into models.
- Design of a better experimental test for modeling may help better
differentiate models.
|
|
22
|
- PID Velocity Control
- PID control performance impaired by varying drive-train dynamics.=
li>
- VSPID controller provided an effective solution to varying drive-tr=
ain
dynamics.
- Supervisory Control
- Successfully detected the state of the plant in both low and high
ratio.
- Improved modeling method could help better differentiate models.
|
|
23
|
- This project is supported by Case New-Holland
|