You are on page 1of 7

FUZZY LOGIC APPLICATIONS TO HIGH PERFORMANCE

DRIVES WITH SPECIAL REFERENCE TO RELUCTANCE


MOTORS
Authorised By
SANTOSH BHARADWAJ REDDY
Email: help@matlabcodes.com
Engineeringpapers.blogspot.com
More Papers and Presentations available on above site

Abstract: every nonlinear control application. In

This paper reviews the the area of variable speed drives and

applications of fuzzy and neuro-fuzzy power electronics, fuzzy logic

control to DC and AC drives. It also applications are rather new.

reviews the application in V.R. drive Nevertheless, they show superior

systems. It is concluded that further performance over the conventional

research work is required in this area controllers in providing the nonlinear

as V.R. drives are quite non-linear control in order to cope with large

electromagnetic structures. Fuzzy and fluctuations in speed and torque in

neuro-fuzzy control techniques are drive systems. Fuzzy logic control has

ideal solutions to control of reluctance already been applied to DC drive

drives. systems, induction motors, permanent

Introduction magnet motors, variable reluctance


motors and disk drives
In recent years, fuzzy logic has
emerged as an important artificial
intelligence tool in order to characterise Fuzzy Logic Control:
and control a system whose model is The theoretical foundations of

not known or it is highly non-linear. It fuzzy logic is available in various

has been widely used in process textbooks and other references and

control, domestic appliances, stock hence not repeated here . The basic

market prediction, machine tools, structure of fuzzy logic control (FLC)

transportation, automotives, elevators, system is shown in Figure 1. The plant

etc. Over the past decade, fuzzy and control signal U is inferred fiom the

neuro-fuzzy control has emerged as a two state variables, error (e) and change

very successful technology in almost in error (ce) which are per unit signals

1
derived from the actual E and CE logic and produces a fuzzy controller
signals. Taking crisp input values of e with the advantages that automatic
and ce and taking the values of the set’s identification of fuzzy rules and
membership function at that input value membership function tuning is
performs the Fuzzification operation. achieved. A typical neuro- fuzzy
The so-called fuzzy rules, are evaluated topology is shown in Figure 2 [23]. It
and then Defuzzification operation is shows that the neural network can be
performed to obtain the resultant crisp interfaced with a conventional fuzzy
control value. The evaluation of rules is logic control. In many applications,
based on a rule-base, which is specified knowledge that describes desired
by the designer. The designer also system behaviour is contained in data
specifies the appropriate membership sets. Hence, the designer has to derive
finction for all the fuzzy sets of each the "IF ---- THEN" rule from the data
variable.. The first step in a fuzzy sets manually which requires a major
control system design is the definition effort with large data sets. When data
of the system structure. The actual sets contain knowledge about the
control strategy of a fuzzy system lies system to be designed, a neural net
in the definition of the rules. promises a solution because it can train
itself from the data sets. Neural nets can
learn from data sets, while fuzzy logic
solutions are easy to verify and
optimise.

Application to Drive Systems:


DC Drives:
DC drives are still commonly
employed in domestic appliances and
high precision control systems. Fuzzy
logic control has been applied in
various DC drive systems P I .
Neuro-Fuzzy Control:
Neuro-fuzzy control applies
neural network techniques to fuzzy

2
passes through the normalising factor
&, and discrete integrator to provide
iref. The fuzzy block performs the
following:
(i) eN and ceN are fuzzified by assigning
them a single fuzzy set.
(ii) The fuzzy set of the output variable
is inferred by fuzzy rule evaluation.
(iii)A crisp value of cref,N is obtained
by defuzzifying the fuzzy output
variable.

The scaling factors play an


important role in the FLC performance.
They can be adjusted by trial and error
to give desired speed (e.g. under
damped type of response). "he fuzzy
logic controller is implemented by
using NN as shown in Figure 5, with
two input neurons, three hidden
neurons and one output neuron. The
input neurons have a linear transfer
The error and change of error are
function and hidden neurons utilizes
calculated first, they are normalised by
f(u) = tanh (U) as transfer function.
scaling factors K, and &. The fuzzy
block does the fuzzy logic processing
and this provides the output as change
in the reference current, which in turn

3
Figure 6. The inputs to the neural
network are:
1) Reference torque
2) Reference speed
3) Reference frame angle and
4) Reference 3-phase currents.

The output is optional switching


signals of each phase current, which in
turn, is fed to the inverter for turning
any phase on and off. The learning
criteria used for neural network are
The training of the NN involves
such as to maintain a constant
by first generating some input pattems
switching frequency. The respective
and by calculating the corresponding
fuzzy current controller structure is
FLC output in Figure 4. This set of
shown in Figure 7. The inputs to fuzzy
input-output data is presented to NN
current controller are the phase current
and by using back proptigation a!
errors e, and change of current errors
gorithm, the weights are computed. The
(e). The outputs are again switching
various experimental results show a
signals of each phase current.
very good performance of:
a) No load response of the drive to a
reference step
b) Step response with high moment of
inertia and friction coefficient and
c) Sudden application of rated torque.

Induction Motor Drives:


A current controller for
induction motor based on neural
network and fuzzy logic theory is
proposed. Firstly, the neuro current
controller has the structure shown in

4
machines employ rare-earth magnets to
reduce the size and improve the
efficiency performanceFuzzy and
Neuro-fuzzy control techniques have
also been applied to control PM drive
systems.
Variable reluctance machines
are robust and well suited for very high
speed applications. These machines are
ideal candidates as generators for wind
energy conversion systems, aerospace
systems and automotive applications
The fuzzy logic block implements the
Combining the neuro controller and
speed regulation. The overall block
fuzzy controller, the neuro-fuzzy
diagram is as shown in Figure 10.
controller has the architecture as shown
in Figure 8.
In order to obtain optional
excitation parameters θ on,θ off and Ich
the following state space model is
used .

Permanent Magnet AC Drives:


Permanent magnet machines are
finding wide applications both in high
precision control and as industrial drive
with high efficiency. Modem PM
5
PM machines where flux-weakening
and θ Off is determined by setting the mode operation is rather difficult at
derivative of torque with respect to θ Off
high speeds.
to zero. The neural network is trained
by representing it with a sequence of Conclusion:
inputs (T, w) and outputs (θ on,θ off , Ich This paper provides a brief
) patterns obtained from the above state description of the neuro fuzzy logic
space model. The fuzzy logic controller control techniques applied to DC and
regulates the speed in the vicinity of an AC drive systems. With the advent of
optimal excitation operating points and fast digital systems, the neuro-fuzzy
to provide good damping near the controllers will become more and more
reference speed. This fuzzy logic practical. Although, these techniques
controller allows the drive system to are increasingly being used in DC,
track large changes in set point induction and PM drive systems but
(reference speed) or load torque. few applications have been reported for
VR drive systems and none for
synchronous reluctance motors. It is
hoped that it is an ideal application area
due to highly non-linear nature of such
drive systems.

References:
1. D. Driankov, H. Hellendoom
and M. Reinfiank, “An
Introduction to Fuzzy
Control”.
2. P. Guillemin, “Fuzzy logic
applied to motor control”
Synchronous reluctance or singly
salient VR motors are robust and can be Engineeringpapers.blogspot.com
operated both in open loop and closed More Papers and
loop. In closed loop, it has the Presentations available on
advantage of wide speed range, unlike
above site

6
7

You might also like