Intelligent and Precision Control Laboratory

FACULTY: Prof. Bin YaoEmail: byao@ecn.purdue.edu

Phone: (765) 494-7746http://widget.ecn.purdue.edu/~byao

Current Graduate Students:

D. Agarwal, J. Chuanasa, R.G. Dontha, P. Garimella, K.A. Gul, J.Q. Gong, Y. Hong, G. Hu, S. Liu, Y. Xu

Visiting Scholars:

Prof. Hiroshi Fujimoto

Sponsors and Donors:

NSF, Maxtor Inc., Caterpillar Inc., Vickers Inc., Purdue Research Foundation, Purdue Electro-

Hydraulic Research Center, …

Research Focus:

• Developing a general framework for the design of intelligent and yet high precision/performance control algorithms

• Applications to the integrated design of intelligent and precision mechatronic systems

• Nonlinear observer design and neural network learning for virtual sensing, modeling, fault detection, diagnostics, and adaptive fault-tolerant control; Data fusion.

NONLINEAR ADPTIVE ROBUST CONTROL (ARC) THEORY

APPLICATIONS

• Control Performance Oriented Direct Adaptive Robust Control• Learning law and control law are synthesized at the same time solely for reducing output tracking error, which leads to control systems having lower-orders

• Estimation Based Indirect Adaptive Robust Control • Decouple the constructions of control law and learning law so that better estimation model and on-line signal monitoring can be used to obtain accurate parameter and/or function estimates for secondary purposes such as prognostics

• Integrated Direct/Indirect Adaptive Robust Control•Intelligently integrating the fast dynamic compensation actions of direct ARC design with the good estimation process of indirect ARC design to synthesize controllers having both excellent control performance and parameter and function estimates

• Nonlinear Adaptive Robust Observer Design for Virtual Sensing and Low Cost Implementation

• Neural Network Adaptive Robust Control for General Learning

• Repetitive Adaptive Robust Control for Repetitive Tasks

PRECISION MOTION CONTROL OF HIGH-SPEED LINEAR MOTOR DRIVE SYSTEMS

ULTRA-PRECISION MOTION CONTROL OF NANO-POSITIONERS

Seamless integration of the fast reaction to immediate feedback information (e.g., nonlinear high-gain robust control) and the slow learning utilizing large amount of

stored past feedback information that is available in the modern computer based control systems (e.g.,

adaptive control) to maximize the achievable control performance with built-in intelligences

Essence

Good Body and Instinct

Fast Instantaneous Reaction !

Brain Power

Good Learning Ability !

⇓

Current Focus Areas

• Precision Control of Electro-Magnetic Motor Driven Mechanical Systems for Precision Manufacturing

• Machine Tools

• High-Speed Linear Motors

• Laser Micro-Machining Processes

• Hard Disk Drives (HDD)

• …

• Ultra-Precision Control of Piezoelectrical Actuator Driven Mechanical Systems for Nanotechnology

• Control of Nano-positioning Stage

• Dual Stage Ultra-High Density HDDs

• …

• Energy Saving Control of Electro-Hydraulic Actuator Driven Systems with Novel Programmable Valves

• Hydraulic Servo-systems

• Hydraulic Excavators

• …

• Coordinated Control of Multi-DOF Mechanical Systems and Multiple Robots for Factory Automation

• Trajectory Tracking Control of Robot Manipulators

• Motion and Force Control of Robot Manipulators (a) in contact with a stiff surface with unknown stiffness(b) in contact with a rigid surface

• Coordinated Control of Multiple Robot Manipulators

QuadratureDecoders

16-bit D/A (4) Channe ls

16-bit A/DChannels

Digi ta l I /O

Du

al P

ort

Mem

ory

DSP

Se rvo Controller

AT-bus

Y-A

xis

X-Axis

Z-A

xis

486-PC

Matsuura MCS510VSS

affiliated with

Ruth & Joel Spira Laboratories for Electro-Mechanical SystemsRay W. Herrick Laboratories

Rotary Servo Motor

Gear

Leadscrew

Table

Linear Motion

MAGNET ASSEMBLY

CURRENT (I)

FORCE (F=IxBf)COIL ASSEMBLY

FLUX DENSITY (B)

MAGNET ASSEMBLY

COIL ASSEMBLY

CURRENT (I)

FORCE (F)

MAGNETICATTRACTION (Fa)

0 1 2 3 4 5 6 7 8 9 10 11−20

−15

−10

−5

0

5

10

15

20

(b)

IAR

C

0 1 2 3 4 5 6 7 8 9 10 11−20

−15

−10

−5

0

5

10

15

20

(c)

DIA

RC

time (s)

0 1 2 3 4 5 6 7 8 9 10 11−20

−15

−10

−5

0

5

10

15

20

(a)

DA

RC

Tracking error without load in µm

0 1 2 3 4 5 6 7 8 9 10 11−20

−15

−10

−5

0

5

10

15

20

(a)

DA

RC

Tracking error with load in µm

0 1 2 3 4 5 6 7 8 9 10 11−20

−15

−10

−5

0

5

10

15

20

(b)

IAR

C

0 1 2 3 4 5 6 7 8 9 10 11−20

−15

−10

−5

0

5

10

15

20

(c)

DIA

RC

time (s)

0 2 4 6 8 100.04

0.05

0.06

0.07

0.08

0.09

0.1

0.11

0.12

time (s)

θ 1

solid: DARC

dotted: IARC

solid(thick): DIARC

0 2 4 6 8 100.02

0.025

0.03

0.035

0.04

0.045

0.05

0.055

0.06

time (s)

solid: DARC

dotted: IARC

solid(thick): DIARCθ 1

Tracking Errors for Point-to-Point Trajectory 2max max( 12 / sec , 1 /sec)a m v m= =

The above results demonstrate the excellent robust tracking performance of the proposed ARC algorithms – Tracking errors are

mostly within 10 mm with final tracking error around the encoder resolution of 1 mm for both loaded and unloaded cases

The above plots show the accurate parameter estimation capability of the proposed DIARC and IARC – estimates of the inertia load

converge to their actual values for both loaded and unloaded cases

* All results in this section were obtained by Li Xu.

Actual Position IIActual Position I

Contouring Errors

Tracking Errors

Desired Position

Contour tracking

MATLAB/SimulinkReal-Time Workshlp

MLIB MTRACE

Real-Time Interface

Complier

ControlDesk

DS1103

EncoderInterface

A/D, D/A,Digital I/O P

WM

Am

plifi

ers

Main Processor

X-Y Stage (Driven by two linear motors)

Enc

oder

Sin

gals

5X m

ultip

licat

ion

Pentium II 400

A/D,I/O

Encoder Signals

Y-axis

X-a

xis

Servo ControllerHost PC

Hall Sensors

0 0 . 2 0 . 4 0 .6 0 . 8 1 1 . 2 1 .4 1 . 6 1 . 8 2-1 0

- 5

0

5

1 0

Co

nto

uri

ng E

rro

r (µ

m)

T i m e ( s e c )

A R C

0 0 . 2 0 . 4 0 .6 0 . 8 1 1 . 2 1 .4 1 . 6 1 . 8 2-1 0

- 5

0

5

1 0

Co

nto

urin

g E

rro

r (µ

m)

T i m e ( s e c )

D C A R C

Contour tracking performance for a circular path with a radius of 0.1m and a feedrate of 314mm/sec (without load)

COORDINATED CONTROL OF MULTI-DOF MECHANICAL SYSTEMS

0 1 2 3 4 5 6 7−1

0

1

2

3

4

5

6

7

8

9

Input

Sen

sor

Out

put

Differences in hysteresis loops at different frequencies

0.01 hz0.1 hz 1 hz 10 hz 50 hz data6

0 5 10 15 20−0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

(sec)

(V)

0 5 10 15 20−0.2

0

0.2

0.4

0.6

0.8

1

1.2

(sec.)

(V)

Input Response

0 5 10 15 200

0.5

1

1.5

2

2.5

3

3.5

4

(sec.)

(V)

0 5 10 15 200

1

2

3

4

5

6

7

8

9

(sec.)

(V)

The above plots show the drift, the non-linear and hysteresisbehaviors of the response of piezo-actuator driven nano-positioners

Step Responses to Different Input Magnitudes Responses to Sinusoidal Inputs

1.3975 1.398 1.3985 1.399 1.3995 1.4−0.2

0

0.2

0.4

0.6

0.8

1

1.2

ery

Ts−2%

≈ 0.8 msec.

% Overshoot ≈ 14

Experimental Tracking Performance with an augmented pre-filter and lag compensator

Unit: 1 V = 1200 nm

Overall Model of Nano-Positioners

The above results demonstrate the excellent robust contour tracking performance of the proposed ARC algorithms

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2−10

−5

0

5

10

Conto

uring

Error

(µ m

)

Time (sec)

ARC

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2−10

−5

0

5

10

Conto

uring

Error

(µ m

)

Time (sec)

DCARC

Contour tracking performance (with load)

Task Space

( ),

( )p p

f f

r r f qr

r r g q

= = =

( ) 0g q =

Transform the coordinated control problem to the control of MIMO nonlinear systems in

task space