4th Power Electronics, Drive Systems & Technologies Conference (PEDSTC2013), Feb l3-14, 2013, Tehran, Iran
A Hybrid Adaptive Neural-Fuzzy Tuned P.I.
Controller Based Unidirectional Boost P.F.C.
Converter Feeds B.L.D.C. Drive
S. Hr. A. Kaholi
UMPEDAC
U.M.
K. L., Malaysia
kaboli0004@gmail.com
M. Mansouri
Dept. ofElec.Eng.
U.M.
K. L., Malaysia
mh.mansouri@gmail.com
Abstract- The demand for solid state A.e.-D.e. converters to
enhance power quality in the expression of P.F.e. (Power Factor
Correction) impels the proposal of miscellaneous topologies. This
paper presents a hybrid Neural-Fuzzy Controller (N.F.e.) tuned
P.I. controller based Unidirectional Boost P.F.e. converter not
only for regulating the drawn current, minimizing harmonics
pollutions (that deteriorating the system power quality) and
enhancing the circuit eiciency but also for B.L.D.e. (Brush Less
D.e.) machine speed controlling. The paper irst presents a
mathematical model and P.F.e. circuit; besides it depicts the way
of hybrid controlling system evolves the controlling of drives
performance at both transient and steady state conditions. It is
important to be concerned that an integral robust S.S.E.E.
(Steady State Error Eliminator) is paralleled to the system
developing the controlling process wholly. Minimizing A.e.
supply current distortion as close as to unity P.F. over a wide
range of rotational speed is one of the advantages of this
enhanced system. The presented modeled system contains Hybrid
Neural-Fuzzy Controlling units, P.F.e. unit, unidirectional Boost
Converter and B.L.D.e. machine; that simulated MA TLAB
model is proposed and discussed in details. Simulation results
signiicantly depict the impressiveness of the proposed
controlling system in T.H.D. reduction.
Kywords-Hybrid Neural-Fuzzy; P.l; Total Harmonic
Distortion; Power Factor Correction; Brushless D.. motors;Boost
Inverter.
I. INTRODUCTION
Servo robotic applications, dynamic actuators and machine
tools are some conventional domination ield of B.L.D.C.
motors. The motor preferred mechanical and elecrical
characteristics; rising eiciency and minimized inertia
momentum are the major factors make B.L.D.C.
implementation such widespread [I], [3], [4].
Nowadays, emphasizing on the power quality improvement
has been increased gradually causes realizing of presenting and
enhancing new clean power converter topologies, meanwhile
there is an increasing demand for high eiciency
implementation of motor drives results in users looking
forward a favorable performance of conrolling which should
be provided meanwhile loads and motors parameter are varying
during unction. By considering what mentioned above, it is
inferred that there is an intense atractiveness in enhancing
adaptive P.F.C. conroller that results in proposing various non-
978-1-4673-4484-5/13/$31.00 ©2013 IEEE 176
J. Selvaraj
Dept. ofElec.Eng.
UM.
K. L., Malaysia
N.B.A. Rahim
Dept. ofElec.Eng.
UM.
K. L., Malaysia
linear model based adaptive P.F.C. conrolling schemes
dedicated for B.L.D.C. motors [I].
Generally, there are conventional unidirectional converters
as buck, boost, buck-boost and multilevel [9] that supply
variable D.C. power demand, although these converters endure
poor P.F., utility injected hannonics and D.C. voltage link
luctuation regarding to supply requency and voltage
variation. A proper solution for direct A.C.-D.C. conversion to
meet high density of power is prepared by implementing a
single phase active power factor controller. The applied
unidirectional boost converter is constituted rom a closed loop
curent controller, is able to meet electricity authorities and
agencies srict power factor penalty limits besides eliminating
input line current hmonics distortion by applying active
conrolling techniques as P.I.D., Fuzzy Logic, Neural Network,
Adaptive Neural-Fuzzy or hybrid corolling method that lead
to realizing power factor corection [8], [10].
P.I. (Proportional-Integral) controllers are one of the well
known techniques applied in industrial conrolling process
regarding to their ease of implementation and robust
performance over a wide range of operational conditions.
Gains of Proportional and Integral are the factors directly affect
the design of the system; it is why designers spend a lot of time
for optimizing the conroller parameters [8]. In nonlinear Boost
Converter P.F.C. Inverter Fed B.L.D.C. system, linear
conroller as P.I. is practically insuicient especially when a
noble ransient perfonnance is desirable under all operational
conditions. A dynamic gain regulation for Proportional and
Integral is a solution that can solve the mentioned problem
thoroughly by transfering controller to new operating points.
The main obstacle for implementing this dynamic solution is
providing a dynamic tuning process respectively [8]. One of
the well-known tuning methods is the Ziegler-Nichols tuning
formula. The technique is quite simple but is not always proper
when parameters are varying besides this technique may not
meet quick on-line tuning. It is also proved that B.L.D.C.
machines are sensitive to parameters variations and load
disturbances, those predominate hmul impacts in B.L.D.C.
conrolling systems.
A.N.F.I.S. (Adaptive Neural-Fuzzy Inference System) is
named for Neural-uzzy network-based systems. This system is
a type of Neural-network based on Takagi-Sugeno uzzy