AUTO-CALIBRATION AND CONTROL APPLIED TO ELECTRO-HYDRAULIC VALVES

1
AUTO-CALIBRATION AND CONTROL APPLIED TO
ELECTRO-HYDRAULIC VALVES
EXPERIMENTS ON HUSCO BLUE TELEHANDLERAugust 18,
2006

• PATRICK OPDENBOSCH
• Graduate Research Intern
• INCOVA
• [phone]
• patrick.opdenbosch_at_huscointl.com

HUSCO International W239 N218 Pewaukee
Rd. Waukesha, WI 53188-1638
2
MOTIVATION
HUSCOS CONTROL TOPOLOGY
US PATENT 6,732,512 6,718,759
Steady State Mapping (Design)
Inverse Mapping (Control)

• Hierarchical control System controller, pressure
  controller, function controller

HUSCO OPEN LOOP CONTROL FOR EHPVs
3
MOTIVATION
HUSCOS CONTROL TOPOLOGY
US PATENT 6,732,512 6,718,759
Steady State Mapping (Design)
Inverse Mapping (Control)

• Hierarchical control System controller, pressure
  controller, function controller

4
MOTIVATION
Time
Commanded Kv
Actual Kv
Commanded Velocity
Actual Velocity
Time
5
MOTIVATION

• Flow conductance online estimation
• Accuracy
• Computation effort
• Online inverse flow conductance mapping learning
  and control
• Effects by input saturation and time-varying
  dynamics
• Maintain tracking error dynamics stable while
  learning
• Fault diagnostics
• How can the learned mappings be used for fault
  detection

6
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

7
TOPIC REVIEW

• PURDUE PAPERS
• Liu, S. and Yao, B., (2005), Automated modeling
  of cartridge valve flow mapping, in Proc
  IEEE/ASME International Conference on Advanced
  Intelligent Mechatronics pp. 789-794
• Liu, S. and Yao, B., (2005), On-board system
  identification of systems with unknown input
  nonlinearity and system parameters, in Proc ASME
  International Mechanical Engineering Congress and
  Exposition
• Liu, S. and Yao, B., (2005), Sliding mode flow
  rate observer design, in Proc Sixth
  International Conference on Fluid Power
  Transmission and Control pp. 69-73

8
TOPIC REVIEW

• CATERPILLAR PATENTS
• Aardema, J.A. and Koehler, D.W., (1999) System
  and method for controlling an independent
  metering valve, U.S. Patent (5,960,695)
• Aardema, J.A. and Koehler, D.W., (1999) System
  and method for controlling an independent
  metering valve, U.S. Patent (5,947,140)
• Kozaki, T., Ishikawa, H., Yasui, H., et al.,
  (1991) Position control device and automotive
  suspension system employing same, U.S. Patent
  (5,004,264)
• NEW PATENTS
• Reedy, J.T., Cone, R.D., Kloeppel, G.R., et al.,
  (2006) Adaptive position determining system for
  hydraulic cylinder, U.S. Patent (20060064971)
• Du, H., (2006) Hydraulic system health indicator,
  U.S. Patent (7,043,975)
• Wear, J.A., Du, H., Ferkol, G.A., et al., (2006)
  Electrohydraulic control system, U.S. Patent
  (20060095163)

9
TOPIC REVIEW

• CATERPILLAR PATENTS
• 20060064971 Adaptive Position Determining System
  for Hydraulic Cylinder

Limit Switches
10
TOPIC REVIEW
Long-Jang Li, US Patent 5,942,892 (1999)

• CATERPILLAR PATENTS
• 5,004,264 Position Control Device and Automotive
  Suspension System Employing Same

Position Detector
11
TOPIC REVIEW

• CATERPILLAR PATENTS
• 20060095163 Electrohydraulic Control System

Position/Velocity sensor
Adaptive scheme no details found
12
TOPIC REVIEW

• CATERPILLAR PATENTS
• 7,043,975 Hydraulic System Health Indicator

Using Lyapunov stability theory
Health Monitoring using Bulk modulus and other
model-based parameters
(Position/velocity sensor)
Based on pump pressure discharge dynamics or
cylinder head end control pressure
13
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

14
SETUP

• MOTION CONTROL
• Independent coil current control
• SIEMENS controller
• Supply return pressure from ISP

Supply
KSA
KSB
HUSCO Blue Telehandler
KAR
KBR
Return
Boom Function
Boom Function Kinematics
15
SETUP

• MOTION CONTROL
• Independent coil current control
• SIEMENS controller
• Supply return pressure from ISP

HUSCO Blue Telehandler
PS
PB
PA
PR
16
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

17
IMPROVEMENTS

• PUMP CONTROL

Pressure override for pump pressure control (ISP
code)
Ripples
18
IMPROVEMENTS
DATA SHOWN Margin added on retract metering mode
(PB signal is user commanded, not actual
workport pressure)

• PUMP CONTROL

Current override for unloader coil current
control (ISP code)
19
IMPROVEMENTS

• PUMP CONTROL

Current override for unloader coil current
control (ISP code)
20
IMPROVEMENTS

• ANTI-CAVITATION

KOUT_MAX
m R3/4
PIN_MIN
Unconstrained Operating Point
Keq_dPmin
KIN_MAX
Keq
POUT_MAX
Constrained Operating Point
21
IMPROVEMENTS

• ANTI-CAVITATION

Cavitation
22
IMPROVEMENTS

• ANTI-CAVITATION

Flow Sharing
No Cavitation
23
IMPROVEMENTS

• LEARNING

Supply
KSA
KSB
EXTEND
KAR
KBR
Return
Boom Function
24
IMPROVEMENTS

• LEARNING

Supply
KSA
KSB
RETRACT
KAR
KBR
Return
Boom Function
25
IMPROVEMENTS

• LEARNING

Supply
KSA
KSB
EXTEND/RETRACT
KAR
KBR
Return
Boom Function
26
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

27
MAPPING LEARNING CONTROL

• LEARNING APPLIED TO NONLINEAR SYSTEM
• MAPPING TO BE LEARNED (simplified)

Expected curve shift
28
MAPPING LEARNING CONTROL

• LEARNING APPLIED TO NONLINEAR SYSTEM
• MAPPING TO BE LEARNED (simplified)

Expected curve shift
29
MAPPING LEARNING CONTROL

• LEARNING APPLIED TO NONLINEAR SYSTEM
• CONTROL DESIGN
• Tracking Error
• Error Dynamics

Linear Time Varying System
30
MAPPING LEARNING CONTROL

• LEARNING APPLIED TO NONLINEAR SYSTEM
• CONTROL DESIGN
• Error Dynamics
• Deadbeat Control Law
• Closed loop

31
MAPPING LEARNING CONTROL

• LEARNING APPLIED TO NONLINEAR SYSTEM
• CONTROL DESIGN
• Deadbeat Control Law
• Proposed Control Law

32
MAPPING LEARNING CONTROL
Nominal inverse mapping
Inverse Mapping Correction
icmd
KV
EHPV
Servo
NLPN
dKV
Adaptive Proportional Feedback
Jacobian Controllability Estimation
33
MAPPING LEARNING CONTROL

• LEARNING APPLIED TO NONLINEAR SYSTEM
• CONTROL DESIGN
• Proposed Control Law
• Closed loop

34
MAPPING LEARNING CONTROL

• IDENTIFICATION DESIGN
• Methods
• Least Squares (Recursive)
• Noise rejection
• Poor time varying parameter tracking capabilities
  (add covariance reset and forgetting factor
  dynamic or static)
• New research suggest variable-length moving
  window
• Gradient Based
• Sensitive to noise
• Better time varying parameter tracking
  capabilities
• Gradient step size must be chosen carefully

Identification of time varying parameter for a
linear system
() Jiang, J. and Zhang, Y. (2004), A Novel
Variable-Length Sliding Window Blockwise
Least-Squares Algorithm for Online Estimation of
Time-Varying Parameters, Intl. J. Adaptive Ctrl
Signal Proc., Vol 18, No. 6, pp. 505-521.
35
MAPPING LEARNING CONTROL

• IDENTIFICATION DESIGN
• Approximations
• Previous-point Linearization
• Stack Operator

36
MAPPING LEARNING CONTROL

• IDENTIFICATION DESIGN
• Approximations
• Previous-point Linearization
• Stack Operator Properties

37
MAPPING LEARNING CONTROL

• IDENTIFICATION DESIGN
• Approximations
• Previous-point Linearization
• Stack Operator Properties

38
MAPPING LEARNING CONTROL

• IDENTIFICATION DESIGN
• Approximations
• Previous-point Linearization

39
MAPPING LEARNING CONTROL

• IDENTIFICATION DESIGN
• Approximations
• Previous-point Linearization

How are (dJ,dQ) and (J,Q) related?
40
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

41
EXPERIMENTAL RESULTS
Nominal inverse mapping
icmd
KV
EHPV
Servo
dKV
Every valve uses a generic Table
42
EXPERIMENTAL RESULTS

• PUMP CONTROL MARGIN

43
EXPERIMENTAL RESULTS
44
EXPERIMENTAL RESULTS
45
EXPERIMENTAL RESULTS

• PUMP CONTROL PS_SETPOINT

46
EXPERIMENTAL RESULTS
47
EXPERIMENTAL RESULTS
48
EXPERIMENTAL RESULTS
Nominal inverse mapping
Inverse Mapping Correction
icmd
KV
EHPV
Servo
NLPN
dKV
49
EXPERIMENTAL RESULTS

• PUMP CONTROL MARGIN

50
EXPERIMENTAL RESULTS
51
EXPERIMENTAL RESULTS
52
EXPERIMENTAL RESULTS

• PUMP CONTROL MARGIN

53
EXPERIMENTAL RESULTS
54
EXPERIMENTAL RESULTS
55
EXPERIMENTAL RESULTS

• PUMP CONTROL PS_SETPOINT

56
EXPERIMENTAL RESULTS
57
EXPERIMENTAL RESULTS
58
EXPERIMENTAL RESULTS
Nominal inverse mapping
Inverse Mapping Correction
icmd
KV
EHPV
Servo
NLPN
dKV
FIXED Proportional Feedback
59
EXPERIMENTAL RESULTS

• PUMP CONTROL MARGIN

60
EXPERIMENTAL RESULTS
61
EXPERIMENTAL RESULTS
62
EXPERIMENTAL RESULTS
63
EXPERIMENTAL RESULTS
64
EXPERIMENTAL RESULTS
65
EXPERIMENTAL RESULTS
66
EXPERIMENTAL RESULTS
67
EXPERIMENTAL RESULTS
SHOW LEARNED MAPS
68
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

69
FUTURE WORK

• Improve EHPV performance using adaptive
  proportional feedback
• Study convergence properties of adaptive
  proportional input and its impact on overall
  stability
• Incorporate fault Diagnostics capabilities along
  with mapping learning
• Refine pump controls

70
PRESENTATION OUTLINE

• MOTIVATION
• TOPIC REVIEW
• SETUP
• IMPROVEMENTS
• MAPPING LEARNING CONTROL
• EXPERIMENTAL RESULTS
• FUTURE WORK
• CONCLUSIONS

71
CONCLUSIONS

• The performance of the INCOVA control system
  under Ps_setpoint and margin pump control was
  improved when using mapping learning as oppose to
  using fixed inverse valve opening mapping.
• Satisfactory experimental results were obtained
  on applying feedforward learning and fixed
  proportional control to four (4) EHPVs
• Experimental verification of improved commanded
  velocity achievement using mapping learning was
  presented
• The need for good velocity sensor was observed
  (potential idea for customized sensor was
  presented)

72
CONCLUSIONS

• More refined code (constraints) allowed better
  control
• Unresolved Issues still exist with parameter
  estimation and adaptive proportional control
  portion
• Experimental validation of faster mapping
  learning with proportional feedback in place
  (fixed)
• Learning grid can be fixed based on curve
  shifting behavior

73
QUESTIONS?