Stephen J. DODDS, University of East London

1
SENSORLESS INDUCTION MOTOR DRIVE CONTROL SYSTEM
WITH PRESCRIBED CLOSED-LOOP ROTOR MAGNETIC FLUX
AND SPEED DYNAMICS

• Stephen J. DODDS, University of East London
• Viktor A. UTKIN, Institute of Control Sciencies,
• Russian Academy of Sciences, Moscow
• Jan VITTEK, University of Transport and
  Communications, Zilina

2
BASIC PRINCIPLE
y
nonlinear plant
LINEARISING FUNCTION
u
specified
u
y
nonlinear control law
nonlinear plant
closed-loop system
y
i.e.,
linear and de-coupled closed-loop system with
prescribed dynamics
MOTION SEPARATION
3
EXTENSION TO INDIRECTLY CONTROLLED VARIABLES
nonlinear plant
u
LINEARISING FUNCTION
z
controlled variables
available measurements
z
nonlinear plant
nonlinear control law
u
specified
closed-loop system
y
i.e.,
z
observer
4
MODEL OF MOTOR AND LOAD
expressed in stator-fixed frame
motor torque
rotor magnetic flux linkage
rotor speed
stator currents
stator voltages
stator and rotor resistances
stator, rotor and mutual inductances
5
CONTROL LAW DESIGN
1. SIMPLIFICATION OF CONTROL PROBLEM BY
INNER/OUTER CONTROL LOOP STRUCTURE
w
r
outer-loop sub-plant
Y
w
d
I
master control law
d
slave control law
Y
d
U
outer loop
inner-loop sub-plant
inner loop
w
Y
r
I
observers
6
2. Slave Control Law

• Two options are considered
• A High Gain Proportional Control Law with
  Saturation Limits
• Bang-Bang Control Law Operating in the Sliding
  Mode
• Automatic Start Algorithm bypasses Slave Control
  Law with simple algorithm,which applies maximum
  voltage to one phase until magnetic flux has
  grown sufficiently.

If
then
7
3. MASTER CONTROL LAWindependently controls
rotor speed and magnetic flux norm with first
order dynamics and time constants, T1 and T2
motor equation
linearising functions
desired closed-loop equation
motor equation
mastercontrol law
desired closed-loop equation
8
3. STATE ESTIMATION AND FILTERING3.1. Rotor Flux
Estimator
based on motor equations
eliminate
ROTOR FLUX ESTIMATION ALGORITHM by numerical
integration
flux estimate then given by-
flux component estimates are limited on the basis
that they have zero long-term averages with
9
3.2. Pseudo-Sliding Mode Observer and Angular
Velocity Extractor
motor equation
U
I
-v
For classical sliding -mode observer-
,
slope KI
I (not used directly)
For pseudo sliding -mode observer-
,
angular velocity extractor
10
3.3 Filtering Observers
Rotor angular velocity and load torque observer
Rotor magnetic flux observer
11
OVERALL CONTROL SYSTEM BLOCK DIAGRAM
12
Simulation Results for High-Gain Slave Control Law
13
Simulation Results for Sliding Mode Slave Control
Law
14
Comparison of Simulated System Behaviour with
Ideal Transfer Function for High Gain
Proportional CL
15
Comparison of Simulated System Behaviour with
Ideal Transfer Function for Bang-Bang Slave CL
16
Experiments with Induction Motor
Experimental Bench of East London University,
UK January 2000
Voltages Ualpha v. Ubeta
Currents Ialpha v. Ibeta
40
1
A
V
20
0.5
0
0
-20
-0.5
A
V
-40
-1
-50
0
50
-1
-0.5
0
0.5
1
Flux Links PSIalpha v. PSIbeta
Ang. Velocities Torque v. time
0.1
200
Vs
rad/s, Nm
0.05
100
0
0
-0.05
-100
Vs
time s
-0.1
-200
-0.1
-0.05
0
0.05
0.1
0
0.5
1
1.5
2
17
Experiments with Induction Motor, wd200 rad/s,
T10.5 s
a1) speed up
c1) estim. rotor speed, SM observer
b) Estimated variables from observers
c) Real and ideal rotor speed
a) stator currents and rotor flux
a2) steady state
b2) estim. rotor speed and load torque
c2) real and ideal rotor speed
18
Conclusions and Recommendations

• Forced Dynamic Control introduces a new approach
  to the control of el. drives with induction
  motors, when behaviour of the rotor magnetic flux
  and rotor speed dynamics are precisely defined.
• The experimental results show good agreement with
  the theoretical predictions.
• Further improvement of the Forced Dynamics
  Control can be done with MRAC or SMC based outer
  control loop.