Artificial Neuro Fuzzy Inference System (ANFIS) for Low Voltage Ride Through (LVRT) Improvement of Doubly Fed Induction Generator (DFIG)

1. HASIM NAVSARIWALA
23/03/2017

2. DFIG: DOUBLY FED INDUCTION GENERATOR
 Adjustable-speed induction machine which is widely
used in modern wind power industry.
 It consists of,
• Wound Rotor with no. of turns 2-3 times that of stator
• An AC/DC/AC converter
 Power is fed to both stator as well as rotor from the
grid.
 Stator directly connected to the grid
 Rotor connected to grid through AC/DC/AC converter.
 This back-back converter has two converters where
grid-side converter used to control the DC-link voltage
and machine-side converter used to control power
tracking
2

3. 3
Fig: Doubly Fed Induction Generator

4. ADVANTAGES…
 Less power consumption about 1/3 of the generated
power
 Provides +/- 30% operational speed range
 Low rated current
 Low cost
 Small capacity of converters required,
 High energy
 Flexible power control
 Variable speed operation,
 Controllable power factor
 Improved system efficiency
4

5. LVRT
 Low-voltage ride through (LVRT) is the capability
of electric generators to stay connected in short
periods of lower network voltage.
 DFIGs should provide low voltage ride-through
(LVRT) capability for grid faults resulting in an 85%
voltage drop or even more.
 i.e they should stay connected to the grid during
and after grid faults, contributing to the system
stability.
 Moreover, they should supply reactive power to the
grid in order to support the voltage recovery.
5

6. 6Figure: Low Voltage ride-through standard
set by FERC U.S

7. ANFIS: ADAPTIVE NEURO-FUZZY
INFERENCE SYSTEMS
 A class of adaptive networks that are functionally
equivalent to fuzzy inference systems.
 ANFIS architectures representing both the Sugeno and
Tsukamoto fuzzy models
 It has minimum constraints so very popular
 It is feedforward and piecewise differentiable
7

8. A TWO-INPUT FIRST-ORDER SUGENO FUZZY
MODEL WITH TWO RULES
8

9. 9

10.  Layer 1 (L1): Each node produces the membership
grades of a linguistic label. An example of a
membership function is the generalised bell function:
 where {a, b, c} are the parameters.
 Parameters in that layer are called premise
parameters.
 Layer 2 (L2): Each node calculates the firing strength
of each rule using the min operator. In general, any
other fuzzy AND operation can be used.
CONTINUE…
10

11. CONTINUE…
 Layer 3 (L3): The nodes calculate the ratios of the
rule’s firing strength to the sum of all the rules firing
strength. The result is a normalised firing strength.
 Layer 4 (L4): The nodes compute a parameter
function on the layer 3 output. Parameters in this
layer are called consequent parameters.
 Layer 5 (L5): Normally a single node that aggregates
the overall output as the summation of all incoming
signals.
11

12. ANFIS LEARNING ALGORITHM
 When the premise parameters are fixed, the overall output is a
linear combination of the consequent parameters.
The output f can be written as,
f = (w1x)c11 + (w1y)c12 + w1c10 + (w2x)c21 + (w2y)c22 + w2c20
 A hybrid algorithm adjusts the consequent parameters in a
forward pass and the premise parameters in a backward pass.
 In the forward pass the network inputs propagate forward until
layer 4, where the consequent parameters are identified by the
least-squares method. In the backward pass, the error signals
propagate backwards and the premise parameters are updated
by gradient descent.
12

13. ANFIS FOR DFIG
 To maintain DFIG synchronism with grid, the measured
values of voltage, phase angle and frequency should be
same as reference value.
 The measured DFIG voltages and currents are compared
with reference values, then the error between these two
and change in error are taken as an input to ANFIS
controller.
 Reduces error in rule base action and makes the system
output closer than the reference value.
13

14. ANFIS CONTROLLER DESIGN
 ANFIS is the fusion of neural network with fuzzy
inference system.
 Fuzzy logic is a branch of artificial intelligence,
characterized by fuzzification, defuzzification and
rule base.
 Requires input and output database for training.
Generally, for linear database back propagation
network is used and for nonlinear database
multilayer feed forward neural network is preferred.
14

15. ANFIS model structure Membership function
15

16. CONTINUE…
 The ANFIS-PI controller combines the ANFIS logic to
the conventional PI controller, so as to have online
fine-tuned of the PI gain parameters in accordance
with the variations in system parameters during the
fault.
 ANFIS-PI controller scheme gives better outcomes
compared to the conventional PI controller scheme
and crowbar protection scheme to improve LVRT
capability of the whole wind farm during the fault.
16

17. REFERENCES
 ANFIS-PI Controller based Coordinated Control Scheme of Variable Speed PMSG based
WECS to Improve LVRT Capability of Wind Farm Comprising Fixed Speed SCIG based
WECS - Dinesh Pipalava and Chetan Kotwal - Department of Electrical Engineering,
Government Engineering College, Rajkot-360005, Gujarat, India,
 Protection of DFIG wind turbine using fuzzy logic Control - Mohamed M. Ismail , Ahmed F.
Bendary - Dep of Electrical Power and Machines Faculty of Engineering, Helwan University
Cairo, Egypt
 Modelling and Simulation of ANFIS Controlled Doubly FED Induction Generator Based Wind
Energy System for Performance Enhancement - K. Rebecca Angeline*, Tripura Pidikiti**
and Srinivasa Kishore Babu Yadlapati***
 Maximum Power Tracking of Doubly-Fed Induction Generator using Adaptive Neuro-Fuzzy
Inference System - P. Siva, E. Shanmuga Priya, P. Ajay-D-Vimalraj
 MODELING, ANALYSIS AND OPERATION OF WIND DRIVEN DFIG UNDER
UNBALANCE NETWORK VOLTAGE CONDITIONS: A REVIEW - Debirupa Hore, Runumi
Sarma - Assam Engineering College, Assam, India.
 DFIG Control Scheme of Wind Power Using ANFIS Method in Electrical Power Grid
System - Ramadoni Syahputra and Indah Soesanti Department of Electrical Engineering,
Faculty of Engineering UniversitasMuhammadiyah Yogyakarta, Indonesia
 LVRT : Low Voltage Ride-Through - J. Dirksen; DEWI GmbH, Wilhelmshaven
17
 Research Papers:

18. THANK YOU…
18