E597 Steering System Input Servo Modeling

1
E597 Steering System Input Servo Modeling
Vehicle Validation

• by Jerome Ivan

2
Summary

• Project goals
• Vehicle Data
• Servo description
• Steering servo model
• Potential reasons for inaccurate emulation
• Things gone wrong
• Things that could have been done better
• Will I continue?
• Schedule shortfalls
• Analysis Tools

3
Project Goals

• Take steering response data from steering servo
  disconnected from vehicle
• Take steering response data from vehicle as a
  function of speed steering servo connected
• Develop and validate steering servo model using
  test data bond graph model
• Use data from the vehicle to derive lump sum
  steering system C, I and R using the bond graph
  model

4
Steering Freewheeling Kd 1000 /- 10deg.
5
Steering Freewheeling Kd 0 /- 10deg.
6
Steering Freewheeling / No Steering Wheel Kd
1000 /- 90deg.
7
Steering Freewheeling / No Steering Wheel Kd 0
/- 90deg.
8
Steering Vehicle 0 MPH Kd 1000 /- 90deg.
9
Steering Vehicle 10 MPH Kd 1000 /- 90deg.
10
Servo System Description

• Steering Servo System

11
Servo System Description

• Motor Chain Drive

12
Servo System Measurements (no motor data)

• Ix armature 452mkg-m2 (calculated)
• Coil R 0.4 ohm (measured)
• Coil I 70uH (measured)
• Steering Shaft K 100Nm/rad (measured)
• Bearing R 0.05Nms/rad (estimated)
• Motor produces 2.5N _at_ 54amps
• Motor K .0463Nm/A (calculated)

13
Servo System Measurements (no motor
data)Homemade Motor Curve
14
Steering Servo Model

• Simulink Controller

15
Steering Servo Model Bond Graph
16
20-Sim Servo SimulationFails to emulate ringing.
17
Possible Reasons for Inaccurate Emulations

• 20-Sim PD Controller was an invalid
  representation of the Simulink controller.
• Inaccurate bond graph parameters
• Bandwidth limitations of actual controller not
  found in simulation.
• Bond graph model has infinite current source /
  vehicle had a 30 duty cycle limit or about a
  20Amp limit.

18
Possible Reasons for Inaccurate Emulations

• Bandwidth of actual controller

19
Possible Reasons for Inaccurate Emulations

• Bandwidth of simulation controller

20
Things Gone Wrong

• I was never able to accurately emulate steering
  controller.
• I Could not complete the task of deriving
  lump-sum I, C R for the steering system.
• Not enough time to run the HP Signal Analyzer in
  vehicle (bode plots.)
• No motor data

21
Things That Could Have Been Better

• I Could have shut off Simulink steering
  controller and taken open-loop data in vehicle.
• Used open-loop data to validate bond graph model
  without PD controller.
• I Could have attempted to build a more
  representative discrete controller in 20-Sim

22
Will I continue to use the Bond Graph Method?

• Yes, the bond graph method works well for driving
  the state space equations.
• Everyone in the department uses Simulink and is
  at the mercy of SAE documentation for developing
  models.

23
Schedule Short Falls

• Week 1- (Mar 27-31) run test stand build servo
  bond graph
• Week 2- Validate servo model take data in
  vehicle (could not validate servo model)
• Week 3- Validate lump sum vehicle model finish
  PPT (without accurate servo model, there was no
  sense in trying to infer lump sum steering system
  I, C R.

24
Analysis Tools

• 20-sim
• MathWorks Simulink
• MatLab
• HP Full Bridge Tester
• Norton, R. L, Machine Design An Integrated
  Approach, Prentice-Hall Inc.,1998

25
FIN

• Ford F-250 Roaming Robot NQE