03-10-2016, 11:33 AM
1457444871-mohammadi2014.pdf (Size: 1.96 MB / Downloads: 14)
Abstract—In this paper, a combined vector and direct power
control (CVDPC) is proposed for the rotor side converter (RSC) of
doubly fed induction generators (DFIGs). The control system is
based on a direct current control by selecting appropriate voltage
vectors from a switching table. In fact, the proposed CVDPC enjoys
the benefits of vector control (VC) and direct power control (DPC) in
a compacted control system. Its benefits in comparison with VC
include fast dynamic response, robustness against the machine
parameters variation, lower computation, and simple implementation.
On the other hand, it has benefits in comparison with DPC,
including less harmonic distortion and lower power ripple. An
extensive simulation study, using MATLAB/Simulink, is conducted
on a 9-MW wind farm composed of six 1.5-MW DFIG-based wind
turbines. The performance of the proposed CVDPC method is
compared with both VC and DPC under steady-state and transient
conditions. Simulation results confirm the superiority of the
CVDPC over either VC or DPC.
INTRODUCTION
I N THE past decades, a great increase in electrical power
demand and depletion of natural resources have made
environmental and energy crises. These have led to an increased
need for production of energy from renewable sources so that the
world wind energy production has grown significantly due to
cleanness and renewability. Wind power generation is estimated
to be 10% of the world’s total electricity by the year 2020 and is
expected to be double or more by the year 2040 [1]. Wind
turbines (WTs), which play a main role in wind energy, are
basically divided into fixed and variable-speed technologies.
Variable-speed WTs have been increasingly employed
recently due to several advantages compared with the fixedspeed
technologies, including maximized power capture,
decreased mechanical stresses imposed on the turbine, improved
power quality, and decreased acoustical noise [2]. The variablespeed
technologies can be further subdivided into two types:
synchronous generators with full-scale converters and doubly
fed induction generators (DFIGs) with partial-scale converters.
The DFIG is particularly employed for high-power applications,
due to the lower converters cost and lower power losses. The
DFIG control comprises both the rotor side converter (RSC) and
grid side converter (GSC) controllers so that the RSC controls
stator active and reactive powers and the GSC regulates dc-link
voltage as well as generates an independent reactive power that is
injected into the grid [3].
Vector control (VC) is the most popular method used in the
DFIG-based WTs [4], [5]. Some of the advantages are precise
steady-state performance, less power ripple, and lower converter
switching frequency. However, it has some disadvantages, such
as its dependence on the machine parameters variation due to the
decoupling terms and high online computation owing to the
pulsewidth modulation (PWM) procedure. Moreover, the coef-
ficients of proportional–integral (PI) controllers, in the conventional
VC, must be optimally tuned to ensure the system stability
within the whole operating range and attain sufficient dynamic
response during the transient conditions [6]. This will deteriorate
the transient performance of VC and affect the system stability
within changing operation conditions. In order to overcome the
aforementioned problems, different nonlinear control methods
such as direct torque control/direct power control (DTC/DPC)
have been proposed [7], [8]. The main advantages of DTC/DPC
methods include fast dynamic response, robustness against the
machine parameters variation, reduction in computation, and
simple implementation. However, they have some disadvantages
including significant torque/power ripples due to the high bandwidth of the hysteresis controllers, variable switching frequency
of the converters, and deterioration of the controller
performance during the machine starting and low-speed operations.
Although many modified methods have been presented to
overcome these problems [9]–[11], their drawback is complex
online computation.
In order to enjoy the benefits of VC and DTC, the combined
VC and DTC (CVDPC) method has been applied successfully to
induction motor [12]–[14] and permanent magnet synchronous
motors [15], [16]. However, the CVDPC method has not been
studied appropriately for the DFIG. In this paper, it is focused on
comparison of VC and DPC by looking for similarities between
their principles and searching for a fundamental common basis.
From this common basis, in order to enjoy the benefits of VC and
DPC and to avoid some of the implementation difficulties of
either of two methods, the CVDPC method is proposed for the
RSC of the DFIG. The proposed CVDPC has several advantages
in comparison with VC, including fast dynamic response, robustness
against the machine parameter variations, lower computation,
and simple implementation. On the other hand, it has
benefits in comparison with DPC, including less harmonic
distortion and lower power ripple. The rest of this paper is
organized as follows. In Section II, the VC and DPC methods
are described and the common basis of them is investigated. In
Section III, the proposed control system and its basic idea are
discussed. In Section IV, simulation results are shown, and
finally, in Section V, the conclusion is presented.
II. COMBINED VECTOR AND DIRECT POWER CONTROL
A. Vector Control
VC is the most popular method used in the DFIG-based
WTs. In this method, the stator active and reactive powers are
controlled through the rotor current VC. The current vector is
decomposed into the components of the stator active and
reactive power in synchronous reference frame. This decouples
the active power control from the reactive power control. The
stator active and reactive power references are determined by
the maximum power point tracking (MPPT) strategy and the
grid requirements, respectively. The phase angle of the stator
flux space vector is usually used for the controller synchronization.
However, if the stator flux-oriented frame (SFOF) is
used, the overall performance of VC will be highly dependent
on the accurate estimation of the stator flux position. This can
be a critical problem under the distorted supply voltage
condition or varying machine parameters. Therefore, in this
paper, the stator-voltage-oriented frame (SVOF) is used for the
controller synchronization. In order to extract the synchronization
signal from the stator voltage signal, a simple phaselocked-loop
(PLL) system is used. The stator active and
reactive powers are expressed as PROPOSED CONTROL SYSTEM
A. The Basic Idea
As shown in Section II, there is a direct relationship between
the hysteresis control of the stator active power in DPC and the
rotor direct axis current control in VC. On the other hand, it is
illustrated that the hysteresis control of the stator reactive power
in DPC closely corresponds to the rotor quadrature axis current
control in VC. Owing to these facts, it is possible to propose a
new control system based on the common fundamentals of both
methods by combining the merits of DPC and VC. This way, it
may be possible to provide a control system with a desirable
performance and a rather simpler implementation. Equivalently,
it may be regarded as a new system without some of the
difficulties associated with either DPC or VC.
As was proven, by faster selection of the power electronic
switches status, DPC can provide faster torque/power response.
This, in turn, is due to the use of a predetermined switching table
instead of a much more time consuming PWM procedure. Also,
the use of hysteresis controllers, which provide inputs to the
switching table, contributes to the fast dynamics of DPC. Therefore,
the hysteresis controllers and switching table are good
candidates for the construction of the new control system. In the
indirect VC system, the direct and quadrature axis components of
the rotor current are controlled instead of the stator active and
reactive power.
B. The Control System Structure
The proposed control system in Fig. 4 is divided into a VC part
and a DPC part that are shown on the left side and the right side of
the figure, respectively. As seen, the system uses the - and -axis
hysteresis current controllers similar to those in VC and the
switching table like the one in DPC. The - and -axis current
commands are conventionally generated by the PI power controllers
and compared with their actual values. Of course, the
reference frame transformation is required as in VC. The - and
-axis flags as inputs to the switching table are produced from the
rotor current errors by the hysteresis controllers. The third input
to the switching table determines the sector through which the
rotor flux vector is passing. It is produced by measuring the stator
and rotor currents and the rotor position. The switching table
provides the proper voltage vectors by selecting the status of the
inverter switches, the same as in DPC. The switching table is
shown in Table I, which produces all eight voltage vectors
including zero voltages. The system lacks PI current controllers,
a PWM, and feed-forward terms that are usually available in the
VC systems. In fact, the system is a current VC system by means
of a voltage vector selection of the DPC type.
CONCLUSION
In this work, with considering the structure of VC and DPC, an
innovative combined control structure based on the common
basis of both methods has been presented for the RSC of the
DFIG. The combined system enjoys the current VC approach,
which generates the rotor current components and uses the DPCbased
switching table. The proposed CVDPC method has been
compared with both the VC-based optimized PI controllers and
DPC in terms of simple implementation, acceptable power
ripples, and suitable dynamic response. As a result, the proposed
CVDPC method provides a compromise of the advantages of
two methods.
In the steady-state conditions, the CVDPC has power ripple as
low as that of VC. The ripple is significantly lower in comparison
with that of DPC. Furthermore, an FFT analysis shows that
CVDPC has a suitable THD as low as that of VC, which is less
than that of DPC. In the transient conditions, the CVDPC
responds to the wind speed variations approximately as fast as
DPC, which outperform VC in terms of dynamic response.
Moreover, the CVDPC-like DPC outperforms VC in providing
proper decoupling and robustness against the machine parameters
variation. Consequently, the proposed CVDPC not only
enjoys lower power ripple as good as VC but also keeps high
dynamic response as fast as DPC.