An improved design for grid based PV control systems using modified fuzzy logic with an improved switched capacitor DC-DC converter

In Grid computing, the interconnecting photo-voltaic (PV) control system needs an efficient converter to convert the low direct current (DC) voltage into DC. Design has been improved for the PV, by using a modified fuzzy logic along with an improved switched capacitor (SC) and a DC Converter. An optimized SC DC–DC converter’s performance in steady state for improving the efficiency and the regulation of output for a class belonging to the SC converters that is applied to the PV systems control is proposed. The model has many power converters for efficiently managing the transfer of power of a long cable that has higher electro-magnetic interference. On the basis of the model proposed, this converter provides a suitable DC voltage conversion ratio. The capacitor’s resistive output on the impedance accounts for the purpose of charging and discharging losses and the resistive conduction losses has been considered. This work will further extend this DC–DC converter into method of control for a maximum power point tracking which belongs to a PV system that makes use of the variable temperature as well as the irradiance conditions and the fuzzy logic controller (FLC). The FLC that is proposed is improved by using a stochastic diffusion search technique.

1. ®

An improved design for grid based PV control systems using modified
fuzzy logic with an improved switched capacitor DC-DC converter

3.K. Mohan Kumar! H. Abdul Reuf R. Umamaheswar?

Abstract

Keywords Direct cure (DC) € Poco vola (PV) System « ay logic and maxi wes pola tac

1 Introduction hs type of renew

3. primization algorithms are employed. Inside these tech
ques, the GLC as well asthe SDS are individually used as
mally

menional hybrid methods. Sach techniques are

preter
oalineaiy without requiring a mathematical model. Als,
hing based

aright point of operation and can handle the

characteristics and the systems behaviour The

on FLC has a proces that consist of fuzzficaton, system

complete operation of this PV panel and thir actuation
Here. an SDS technique based on fuzzy-logic has been

the PV system's M
ethos used i the work ar explained. In Sect. 4 the

PT. The remaining inves

resalta ofthe experiment are discus and he conclusion is

2 Related works

The novel batery charger that is based on the PV has been
designed and implemented by Fahabadi 9). The main parts
Converter which includes one sich and two diodes and al
a novel MPPT controller The sole si

the ze

age conditions of switching
nce in switching, This MPPT

to ensure that there ae 4

to entity the MPP. Th
Kot and Shi

lapa proye is bet ane the

[10] made a presentation

a high pee
DC boost is being

«cu

MPP

another ficient technique tat a

as à if deep PY system and ts behaviour

andihe PSC
‘hosen asthe Power C

Seyedmabmoudian e al

‘ofthis GMPP Lay thee els have ben presented and
the partici swarm optimization (PSO) metho which are the

common methods in Hiteratre, Tere are four diferent ac
iors which are the speedo

‘convergence, the efficiency (be
the stably (he educ

lation) andthe cost of computation that have been considered
in comparison with hat ofthe technique of PSO.
Chatrenour e al. [12] made a proposal for an Improved
Double Integral Sliding Mode MPPT Controle ADISMC
hich sused fora stand-alone PV sytem. The Sliding Mode

ave been selected by using the Hurwite stability theorem,

percentage of steady sate emor owing

the use of tacking

the param

‘ter of models of solar panes
nan etal 13 furiber presented another novel tech
nique of alicia intelligence which had been based onthe
Adaptive neuro-fuzzy inference system (ANFIS) The cel
a for the prediction of the duty cycle tat belongs à
the electronic switch ofthis DC-DC conver that has been

amed etal [14] a

Mo

Presented an algorith that was

tated MPPT that has had the capacity fr a high energy

‘controler considera

niques of artificial intelligence
been developed bas capability of regulation of voltage rez
uation fr the lad protection fromthe over vote which

Rezvanietal 15} made a presentation of nove

the purpose of achieving

apology
metre. Each

Arla
another new 5

‘these PV cell have spec points named
o] further designed
optimized Hopfield neural network (NN
nique. This Hopfic NN has een used in

forte tuning ofthe fuzzy function of membership. All these

anand Sutra

‘converter along with a controller of MPP tacking have al
een implemented in 100 of the Maib-Simulink forthe

4. 3.1 Slow switching limit (SSL) capacitor optimization

‘ofthe capacitors and will be proportional 10 that ofthe V-Q

5. ‘Guster Computing

these ripple voltage on each ofthese is observed tobe directly
proportional o hat ofthe voltage rated for that capacitor.

‘The impedence of this optimized output may further be
calculated by a combination of both (7) and (8)

m

@

By means of these types of optimization of these capac-
or all the output impedance is found to be proportional
to hat of the square ofthe sum of products of voltages and
their charge flows (V-A product) ofall the capacitors. This
‘optimization can further improve the SC converter and its
performance which has been duly designed in a manner that
is ad-hoc and in particular having a very large ratio of con-

If all such capacitors belonging tothe SC converter are
being rated for same types of such voltage in these ladder
topologies or in case ofthe applications having that ofthe
integrated capacitors their results of such optimization may
besimpliid. Here the constraint total capacitance init value
(of Cie: and each capacitor value may be depicted by (9):

acıl
Den

Every capacitor will be proportionally sized to the charge of
such multiplier coefficients. Along with the optimized capac:
tors, each ofthe voltage riple is sent in a magnitude hat is
found to be equal
"The output impedance of this optimized
les 1 (10)

Ge Cot o

rom (8))

esults of optimization for technology of single-voltage
lesigns of switched-capacitor converter.

do

3.2 Stochastic diffusion search (SDS)

‘The SDS aso performsetectively a suitable match that exists
between these objects within the search space and also that
OF the description of an objet in target fuer implies
€ target in case exits

can also locate an object with a similar description that Is
available t0 the target. The space as well as the object are
(defined asthe atomic data units (ADU) that has set of asie
very object that is par of this search space and ts

target objects have been duly described in terms this ADU,

and it does not contain any other feature. The ADU is taken
to be single pixels and their intensities in case the search
space is another bit map and image and can also constitute
certain higher types of properties like the horizontal or the
vertical lines, the semicircles and the angles. Each ofthese
agents will act autonomously, and in case they are parallel
to that of he thers that ry to locate a position in which the
target within the same space [20]. The target's position has
also been represented as the coordinates of these reference
point that ae predefined in the description ofthe target. The
transmission as well as thei diffusion of such information
(where the exchange ofthe co-ordinates of such a reference
point and locations that are potential) where the agents have
been enabled to communicate with one another and further
allocate their resources of computation in a dynamic manner
on the search’ outcome. Based on the performance wherein
the search agents may become much more active in case
they point out at right positions inside their search space or
else they continue to remain completely inactive. Al these
agents have access to search space and also 10 that of the
target objects description.

In thee initial stages, all these agents are initialised ran-
domly o a particular point of reference within the search
space (for example if the task given is the location of the
word ABRACADABRA inthe Encyclopaedia Britannica in
which case these agents will indicate the positions of the est

Is assumed hat this word ABRACADABRA

tic check independently forthe reference point information
by means of comparing an ADU which is random from their
target object (for example one agent may choose another et
ler” form the word ABRACADABRA) with thatof its cor.
responding ADU inside the search space (which isthe third

estro the reference points right). Incase this testi proved

successful then the agent gets ative or esc it stays inactive,

ity of any agent effectively is nothing but an ini

y point towards te right position. But

fschecked at

aime) there are high chances of false positives (where there

Ús the activation on words tha are like CURVATURE which
fare non-target words in which the third letter from the

belonging to are R which isthe target that

not be excluded and so is the chance of these falo negatives

activation of

search space fo
‘word ABRSCADABRS: is wrongly spel
Subsequently, in a phase of diffusion, each ofthese inac-
tive agent will make a choice of a random one another agent
forthe purpose of communication. Base on whe

EHER

7. D(Y=D(-1)+AD
End

elself AP <0
1rA1>0
DQ) = Dal) -
AD

Else
D(D=D(t-1)+

and Al is NB then AD ls NS
'S and Al is PB then AD is NS

ICAP is NB and Al is PB then AD is NB
ICAP is NB and Al is PS then AD is NB
ICAP is NB and Al is NB then AD is PB.
ICAP is NB and Al is NS then AD is PB
ICAP is PS and Al is PB then AD is NS.
ICAP is PS and Al is PS then AD is
ICAP is PS and Al is NB then AD is Ps
Ir AP is PS and Al ls PB then AD Is PS

8. IL à |

9. PV output power W-Fuzzy
PV output poser W- Proposed Fuzzy

5 Conclusion