Overvoltage Suppression Strategy for VSG-Based DFIGs Under Commutation Failures of HVDC Transmission Systems

Abstract

Virtual synchronous generator (VSG) control, which can provide inertia output, damp power oscillations, and offer frequency and voltage support to power grids, has become a growing trend in the control field of wind power generation. As a new technology, there are still challenges that VSG control has not solved well, such as transient overvoltage suppression. A kind of transient overvoltage, which often occurs during the commutation failures of HVDC transmission systems, will trigger a mass of wind turbine generators (WTGs) disconnecting from grids. To reduce the grid-disconnection risk of the virtual synchronous generator control-based doubly fed induction generators (VSG-DFIGs), this paper first analyzes the mechanism of the automatic voltage regulation (AVR) control usually employed by VSG-DFIGs, then proposes measures to suppress the transient overvoltage. To solve the problem of the reactive power response lag issued by VSG-DFIGs, which will further aggravate the transient overvoltage in continuous low and high voltage faults, the time constant of the AVR control is switched. To fully exploit the potential of the DFIGs’ reactive power support, the droop coefficient of the AVR control is switched during the abnormal voltage stages. The switched droop coefficient will change the rotor excitation current magnitude, thus adjusting the internal potential of a DFIG, finally better supporting or suppressing the terminal voltage during the low or high voltage periods. Simulation results based on the DIgSILENT/PowerFactory platform demonstrate that the proposed method can effectively suppress the transient overvoltage that occurs in continuous low and high voltage events caused by the commutation failures of HVDC transmission systems, reducing the number of WTGs disconnecting from the grids.

1. Introduction

With the continuous decrease of costs and the maturation of technology, the penetration of wind power generation in power systems is constantly increasing [1]. Up to now, most wind turbine generators (WTGs) in service, such as doubly fed induction generators (DFIGs), are controlled by the power decoupling controls based on phase-locked loops (PLLs). This typical grid-following control makes WTGs miss the inertia support, damping, and flexibility of voltage and frequency regulation that traditional synchronous generators can provide [2]. Virtual synchronous generator (VSG) control technology, as a mainstream of grid-forming control, can provide inertia output, damp power oscillations, and support the frequency and voltage of power grids, which has become a trend in the control field of new energy generation.

Fault ride-through capability is crucial to ensure the stable grid-connected operation of WTGs [3,4]. Although the fault ride-through research for the traditional grid-following WTGs is relatively mature, the original control strategies for the grid-following WTGs are no longer applicable to the VSG-based WTGs, since the fundamental control structure changes when the VSG control strategies are applied to them [5,6,7]. Therefore, it is necessary to design new control strategies to ensure the stable operation of the VSG-based WTGs, such as virtual synchronous generator control-based doubly fed induction generators (VSG-DFIGs), when they suffer voltage faults.

Different fault ride-through approaches have been proposed for VSG-DFIGs in the context that the control structure of VSG-DFIGs is already different from that of DFIGs based on PLLs. In the event of an external voltage dip or surge, the primary components that must be suppressed are the transient components [8]. When a symmetrical fault occurs in the grid, the output voltage of the rotor side converter (RSC) varies according to the electromechanical time constant to respond to the grid voltage change, making it difficult to quickly track the voltage fault component [9]. Reference [10] proposed a voltage compensation strategy to overcome the shortcomings that existing VSG control strategies ignore in the transient process of the virtual excitation current. By controlling the RSC to generate a corresponding transient output voltage component, the effect of the rotor transient counter potential on the rotor winding overcurrent can be weakened or even canceled. When an asymmetrical drop or rise in external voltage occurs, the grid voltage can be decomposed into positive-sequence, negative-sequence, and zero-sequence components according to the symmetrical component method, and different suppression voltage components can be adopted for each fault component. Reference [11], on the basis of reference [10], utilized the symmetrical component method to improve the transient transition process for asymmetrical faults in power grids. The transient components of the rotor reaction potential are given through calculation, then the feed-forward compensation control of each fault component is carried out respectively. In the meantime, the compensation of the fault component is limited to ensure that the DC bus hardware protection device does not operate and that overmodulation does not happen to the RSC. However, the calculation of transient voltage is complicated, and the fault ride-through control strategy can be made easier if the virtual resistance control strategy is adopted. In reference [12], the virtual resistance control is equivalent to introducing a rotor transient compensation voltage on the same time scale as the rotor overcurrent at the instant of fault. The rotor transient reverses’ electromotive force is suppressed, which reduces the rotor overcurrent during the fault and thus protects the RSC.

The above strategies all target single voltage step-down or step-up faults. At present, wind energy resources and load centers in many countries are geographically far away from each other, and abundant wind power is transported to the load centers through high voltage direct current (HVDC) transmission systems for large-scale and long-distance transmission. When a fault occurs in the receiving-end grid of a HVDC transmission system, it will cause a commutation failure in the receiving-end converter. This makes the voltage magnitude of the sending-end grid present a continuous change of “first low and then high”. The DFIGs close to the sending-end converter, whose stators are directly connected to the grid, are very easy to fail or cross into failure. Furthermore, most wind farms will contribute reactive power to help maintain the grid voltage during a period of low voltage ride-through (LVRT). However, there is a delay in the reactive power response of WTGs due to factors such as signal acquisition, decision-making, and control [13]. This delay results in the reactive power not being withdrawn on time at the moment when the grid’s low voltage disappears, which will further worsen the transient overvoltage of the sending-end grid in the events of commutation failures. If the transient overvoltage exceeds the threshold value, WTGs will be triggered to disconnect from the grid or even chain collapse will happen to the system [14]. As the proportion of wind power in the grid increases, to avoid a number of WTGs disconnecting from the grid during commutation failures, an HVDC transmission system usually operates with reduced power. In other words, the power of the HVDC transmission system in actual operation is usually lower than its rated transmission capacity. This leads to a situation where the higher wind power output, the lower the actual power of the HVDC transmission system, which makes the capacity of the HVDC transmission system not fully exploited.

Therefore, it is important to design suitable control strategies for the WTGs to suppress the transient overvoltage caused by commutation failures. The stator magnetic chain and the electric potential of a DFIG during commutation failures were analyzed and studied in reference [15]. To clarify the dynamic characteristics of a DFIG during the continuous change of terminal voltage, reference [16], based on the transient reactive power characteristics of DFIGs, proposed an improved control strategy to suppress the overvoltage of the sending-end grid under commutation failures through the voltage detection delay compensation of the DFIG. Reference [17] proposed a control strategy of using a rotor current reference value to reversely track the stator current, which accelerated the decay rate of the transient magnetic chain while suppressing the electromagnetic torque pulsation. The dynamic process of the system was thus shortened.

To summarize, the mentioned grid fault response strategies have the following major drawbacks. Firstly, the calculation process of transient voltage is complicated, which not only increases the difficulty of control implementation, but also may lead to response delays and affect the control effect. Secondly, these strategies mainly target a single voltage step-up or step-down change, so their adaptability and effectiveness are limited for the more complex and variable voltage changes that may be encountered in the actual grid. For example, after a commutation failure in the HVDC transmission system, the rapid transient process of the voltage presenting “first low and then high” challenges the fast response ability of the control strategy. Finally, although the dynamic process of transient voltage under HVDC commutation failure is taken into account, most of the existing strategies for suppressing transient overvoltage are specific to the traditional power decoupling control based on phase-locked loops (PLLs). It is hard for these methods to directly applied to the VSG control.

To address the shortcomings of the above studies that do not simultaneously consider the structural change of VSG-DFIG and the dynamic change of the voltage of the sending grid under commutation failure, this paper proposes an overvoltage suppression strategy for VSG-DFIGs applicable to the case of commutation failures in HVDC transmission systems. The overvoltage suppression strategy is based on the VSG-DFIG structure proposed by our research team [18]: the rotor current-induced electromotive force (EMF) is adopted as the internal potential of a DFIG to realize the VSG control, which can make the DFIG free from the traditional PLL structure to connect to the grid and improve its grid stability; the automatic voltage regulator (AVR) control is adopted to realize the voltage support to the grid. On the basis of this VSG control with the AVR control, the response time of fault ride-through can be shortened by switching the time constant of the AVR control link in the VSG-DFIG to reduce the hysteresis of the rotor excitation current and reactive power. The switched droop coefficient of the AVR control link can fully exploit the DFIGs’ potential to absorb excess reactive power as much as possible to suppress the transient overvoltage caused by the commutation failures of HVDC transmission systems. The proposed strategy is verified by the DIgSILENT/Power Factory platform [19]. These simulation results show that the proposed method can effectively suppress the transient overvoltage which occurs in continuous low and high voltage events caused by the commutation failures of HVDC transmission systems, and then reduce the number of WTGs disconnecting from the grids. This will help a HVDC transmission system fully exploit its capability or increase the proportion of new energy generation in the area close to its rectifier station.

2. The Basic Control of VSG-DFIG and the Transient Voltage Characteristics of HVDC Transmission Systems

2.1. The Basic Control of VSG-DFIG with AVR

According to the motor convention, the T-equivalent circuit diagram of the stator and rotor of a DFIG in the dq coordinate system is shown in Figure 1.

According to the T-equivalent circuit diagram of the DFIG, the expression for the stator terminal voltage is as follows:

$$U_{\mathrm{s}}=R_s{I}_s+j{\omega }_g{\psi }_s+\mathrm{p}{\psi }_s$$

(1)

where U_s is the stator voltage; R_s is the stator resistance; p is the differential operator; Ψ_s is the stator magnetic chain; and ω_g is the angular velocity of the grid.

Neglecting the change in the stator magnetic chain, the stator voltage can be expressed as:

$$\begin{array}{l}{U}_{\mathrm{s}}=R_s{I}_s+j{\omega }_g{\psi }_s\\ =R_s{I}_s+j{\omega }_g{L}_m{I}_s+L_{s\sigma }{I}_s+j{\omega }_g{L}_m{I}_r\\ =(R_s+j{\omega }_g{L}_{\mathrm{s}})I_s+j{\omega }_g{L}_m{I}_r\end{array}$$

(2)

where L_m is the stator and rotor’s mutual inductance; L_sσ is the stator leakage inductance; I_r is the rotor current; and L_s is the stator inductance.

The electromotive force (EMF) generated by the rotor current is designated as the virtual internal potential. To incorporate the kinematic properties of a synchronous machine into this internal potential, the rotor current vector is aligned with the d-axis of the coordinate framework established by the virtual rotor movement [18].

Consequently, the formulas for virtual internal voltage and equivalent internal resistance can be expressed as follows:

$$E=j{\omega }_g{L}_m{I}_r$$

(3)

$$Z=R_s+j{\omega }_g{L}_{\mathrm{s}}$$

(4)

The control refers to the vector control structure of the grid-following DFIGs. In this structure, the fundamental rotor current regulation stays the same, however, the system of control coordinates is transitioned from the PLL’s coordinate framework to the virtual rotor coordinate framework. Additionally, the control object shifts from managing the d-axis active current and the q-axis reactive current to regulate the rotor currents that are anchored on the d-axis of the coordinate framework. The voltage vector diagram for the DFIG under this regulation is illustrated in Figure 2.

When the grid voltage fluctuates, the DFIGs need to provide reactive power compensation to keep the terminal voltage stable. This control strategy imitates the automatic voltage regulation control of traditional synchronous machines [18]. It uses voltage droop control to adjust the rotor excitation current, thereby maintaining the terminal voltage magnitude. The control block diagram is illustrated in Figure 3.

The automatic voltage regulation control can be expressed as:

$$\Delta I_{r0}=K_u(U_{ref}-U)$$

(5)

$$I_{r\_ref}=I_{r0}+{\displaystyle \frac{K_{\mathrm{u}}}{{\tau }_{e}s+1}}(U_{ref}-U)$$

(6)

where I_r0 is the reference value of rotor excitation current; K_u is the voltage droop coefficient; τ_e is the response time constant of the automatic voltage regulation control; and U_ref is the reference value of the stator voltage.

The overall control strategy block diagram of the DFIG’s RSC is shown in Figure 4.

2.2. The Transient Voltage Characteristic of the Sending-End Grid Under the Commutation Failures of HVDC Transmission Systems

From the commutation principle of HVDC transmission systems [20], it can be seen that the relationship between the inverter-side over-trigger angle β, the turn-off angle γ, the phase-change overlap angle μ, and the inverter-side DC voltage U_di are as follows:

$$\beta =\mu +\gamma$$

(7)

$$\gamma =\arccos [({\displaystyle \frac{U_{di}}{N}}+{\displaystyle \frac{3}{\pi }}{X}_i{I}_d)/(1.35U_2)]$$

(8)

$$\mu =\arccos [({\displaystyle \frac{U_{di}}{N}}-{\displaystyle \frac{3}{\pi }}{X}_i{I}_d)/(1.35U_2)]-\gamma$$

(9)

where β, γ is the inverter side over-trigger angle, shutdown angle; μ is the phase-change overlap angle; X_i is the phase-change reactance of the inverter side; I_d is the current of the DC line; and U₂ is the AC bus voltage of the inverter side.

From this formula, it can be seen that when the AC voltage on the inverter side U₂ decreases, U_di decreases, and the DC current I_d rises so the overlap angle μ increases and the shut-off angle γ decreases. When the turn-off angle is less than the minimum, due to the positive voltage, the valve in the converter valve that is supposed to turn off will conduct again before fully recovering its blocking ability. Both upper and lower valve groups of a given phase will conduct simultaneously, causing a short circuit in the bridge arm, which in turn leads to the DC voltage of the receiving-end converter dropping to zero.

After a commutation failure in an HVDC transmission system occurs, the bus voltage at the sending end will present a transient process of “first low and then high”. The reactive power imbalance at the sending end of the system during the fault is the root cause of the transient overvoltage.

When a commutation failure occurs in the inverter-side converter, the DC voltage at the receiving end drops rapidly, the DC current rises, and the sending-end converter current control initiates the voltage dependent current order limiter (VDCOL) control to increase the trigger angle for inhibiting the current rise, which causes the reactive power consumption of the sending-side converter to increase rapidly. The reactive power is supplied to the converter from the sending-end grid, which results in a voltage drop in the rectifier bus.

When the trigger angle starts to rise, I_d still keeps rising. It reaches the peak value at the end of the action delay. The trigger angle starts to decrease during the recovery period, and then the fault enters the DC current reduction stage. After I_d decreases, the reactive power consumed by the rectifier decreases, but the sending-side filter still provides reactive power. This results in a significant excess of reactive power being sent back to the sending grid, leading to temporary overvoltage at the rectifier bus.

3. Design of the Overvoltage Suppression Strategy

To reduce the risk of WTGs disconnecting from grids, this paper researched how to make VSG-DFIGs with AVR better suppress the transient overvoltage caused by commutation failures in HVDC transmission systems. The focus was on the improvement of the continuous low and high voltage ride-through control strategies for VSG-DFIGs. The actions of hardware protection devices such as the Crowbar and the Chopper during the fault period are not discussed in this study.

Due to the characteristics of the sending-end grid voltage magnitude under commutation failures, decreasing first and then increasing, it was obvious that the continuous low-high voltage ride-through capabilities were required for WTGs. A typical requirement of the wind power grid-connection guidelines is shown in Figure 5 [21]. When the grid connection point’s voltage at the wind farm varies in the shaded portion, sandwiched between the two red lines, it is required that the WTGs in the wind farms remain in grid-connected operation. There are currently no clear regulations regarding the duration of the low-pressure phase, the transition phase, and the high-pressure phase.

In this research, the response time and rotor excitation characteristics of the DFIGs were enhanced and upgraded in accordance with the wind power grid integration guidelines to ensure their rapid and active support for the external grid voltage during faults.

3.1. Time Constant Switching

In a conventional synchronous generators’ excitation system, the excitation system can be grouped into the fast excitation system and the slow excitation system according to the magnitude of the time constant τ_e. The former system generally meets the time constant τ_e ≤ 0.1 s; the time constant of the latter varies depending on the electromechanical factors and the type of excitation, with a general range of 0.2 s ≤ τ_e ≤ 1 s [22]. For conventional synchronous generators, the appropriate excitation system is selected according to the system requirements and cannot be adjusted.

However, the VSG-DFIG control strategy adopted in this paper incorporated an auto voltage regulation control link that imitates a conventional synchronous machine; the difference was that the droop coefficients and time constants can be switched according to the system state.

The AVR control was switched from a slow excitation system to a fast excitation system by changing its response time constant, which can accelerate the dynamic reactive power output of the VSG-DFIG and respond quickly to grid faults.

The time constant of the AVR control link was 0.5 s when the DFIG did not enter the fault ride-through stage, for the sake of the steady state of the DFIG during normal operation; the time constant was reduced after the DFIG entered the fault ride-through as a way to reduce the lag in the reactive power produced/absorbed by the DFIG.

The control expression is:

$${\tau }_e=\left\{\begin{array}{l}0.5\hfill \\ 0.01\hfill \end{array}\right. \begin{array}{r}\hfill 0.9<U<1.1\\ \hfill U\le 0.9 or U\ge 1.1\end{array}$$

(10)

The dynamic reactive power support of the VSG-DFIG can be greatly accelerated by scaling down the AVR time constant during voltage dip faults.

3.2. Droop Coefficient Switching

The stator reactive power equation can be given by combining the voltage-vector relationship of the DFIG shown in Figure 2:

$$Q_{\mathrm{s}}=-U_{sq}{I}_{sd}-U_{sd}{I}_{sq}={\displaystyle \frac{E{U}_{\mathrm{s}}\sin \sigma }{X_s}}$$

(11)

Neglecting the change in the power-angle σ during the transient process:

$$Q_{\mathrm{s}0}={\displaystyle \frac{E_0{U}_{\mathrm{s}0}\sin \sigma }{X_s}}$$

(12)

$$Q_{\mathrm{s}1}={\displaystyle \frac{(E_0+\Delta E)(U_{\mathrm{s}0}+\Delta U)\sin \sigma }{X_s}}$$

(13)

Subtract (13) from (12):

$$\Delta Q={\displaystyle \frac{(E_0\Delta U+\Delta E{U}_{\mathrm{s}0}+\Delta U^2)\sin \sigma }{X_s}}$$

(14)

where Q_s is the stator reactive power of the DFIG; Q_s1 is stator reactive power of the DFIG during the transient period; Q_s0, E₀, U_S0 are the stator reactive power, the internal EMF, and stator voltage during the steady state period; and ΔQ, ΔE, ΔU are the deviations of stator reactive power, the internal EMF, and stator voltage during the fault period.

According to the relationship between Equation (5) and the internal EMF and rotor current in the automatic voltage regulation control, it can be seen:

$$\Delta E=X_{\mathrm{m}}\Delta I_{r\_0}=X_{\mathrm{m}}{K}_u\Delta U$$

(15)

Bringing Equation (14) into Equation (13):

$$\begin{array}{l}\Delta Q_{\mathrm{s}}={\displaystyle \frac{(E_0\Delta U+X_{\mathrm{m}}{K}_u\Delta U{U}_{\mathrm{s}0}+\Delta U^2)\sin \sigma }{X_s}}\\ ={\displaystyle \frac{[(E_0+X_{\mathrm{m}}{K}_u{U}_{\mathrm{s}0})\Delta U+\Delta U^2]\sin \sigma }{X_s}}\end{array}$$

(16)

Neglecting the quadratic term of the value of the voltage changes ΔU², there is:

$$\Delta Q_{\mathrm{s}}\approx {\displaystyle \frac{(E_0+X_{\mathrm{m}}{K}_u{U}_{\mathrm{s}0})\sin \sigma }{X_s}}\Delta U$$

(17)

Under the automatic voltage regulation control link, the reactive power ΔQ_s added/absorbed by the stator of the DFIG is approximately linearly related to the value of the voltage variation ΔU. Therefore, by changing the droop coefficient K_u, shown in Figure 4, during the terminal voltage fluctuations, more reactive power can be added/absorbed for the same magnitude of voltage dips/surges, then the support/suppression of the DFIGs’ terminal voltage is better.

To improve the virtual excitation characteristics during the low/high voltage ride-through and then better support or suppress the transient voltage, the principle of modification of the droop factor K_u in the AVR control link is as follows:

$$K_{\mathrm{u}}=\left\{\begin{array}{c}\begin{array}{cc}{\displaystyle \frac{I_{r\mathrm{m}\mathrm{a}\mathrm{x}}-I_{r0}}{0.2-0.9}}& 0.2\le U\le 0.9\end{array}\\ \begin{array}{cc}{K}_{u\_\mathrm{normal}}& 0.9<U<1.1\end{array}\\ \begin{array}{cc}{\displaystyle \frac{I_{r0}-I_{r\mathrm{m}\mathrm{i}\mathrm{n}}}{1.1-1.3}}& 1.1\le U\le 1.3\end{array}\end{array}\right.$$

(18)

where K_{u_normal} is the slope of the excitation curve of the DFIG during steady state operation; I_rmax is the maximum value of the DFIG’s rotor excitation current; I_rmin is the minimum value of the DFIG’s rotor excitation current; and I_r0 is the value of rotor excitation current of the DFIG when its terminal voltage is the rated voltage.

Based on the above expression for the improved voltage ride-through excitation characteristics, the graph of the excitation characteristics of the VSG-DFIG during the low and high voltage ride-through and during the steady state is shown in Figure 6.

The discontinuity of the excitation control curve hardly affected the operation of DFIGs because the control, switched from the steady state operation to the low-or high-voltage ride-through operation, was also a transient process. The improvement of the excitation characteristics of the VSG-DFIG during low- and high-voltage ride-through, as described above, enabled DFIGs to respond quickly to voltage dips/surges in the grid during faults and to provide fast dynamic reactive power support to the grid.

4. Simulation Verification

In order to verify the effectiveness of the proposed control strategy, a wind power delivery system via an HVDC transmission system was constructed in the DIgSILENT/PowerFactory simulation platform [19], in which the DFIGs were modeled as the aforementioned VSG control with an AVR. The topology of the simulation system is shown in Figure 7:

The HVDC transmission system parameters are listed in

Table 1. HVDC system parameters.

Parameters of the HVDC System	Value
Rated transmission capacity Sdn	1000 MW
DC rated voltage Udc	500 kV
Sending-side grid SCR	1.65
Sending/Receiving-side bus voltage	345 kV/230 kV

, the VSG control parameters are listed in

Table 2. Virtual rotor control parameters.

Parameters of VSG	Value
Moment of inertia Jv	400 kg·m2
Damping factor D	950 Nms/rad
Proportional gain Kp	0.5
Integral gain Ki	80

and the DFIG parameters are listed in

Table 3. Parameters of the 5 MW doubly fed induction generator.

Parameters of DFIG	Value
Rated power Sn	5 MW
Rated voltage Us	0.69 kV
Stator reactance Xs	0.08 p.u.
Stator resistance Rs	0.01 p.u.
Mutual inductive reactance Xm	2.7 p.u.
Rotor reactance Xr	0.072 p.u.
Rotor resistance Rr	0.01 p.u.
Rated frequency fn	50 Hz

.

Four simulation scenarios were set up and the details are listed in the

Table 4. Simulation scenarios.

Case	Short-Circuit Location	Short-Circuit Type
1	①	three-phase short-circuit
2	①	two-phase short-circuit
3	①	single-phase short-circuit
4	②	three-phase short-circuit

.

In Table 4, the short-circuit position ① means that the short-circuit occurs in the inverter side bus; the short-circuit position ② means that the short-circuit occurs in the connection line between the inverter station and the receiving end grid.

A comparison of the terminal voltages of the VSG-DFIG before and after adopting the strategy is shown below.

Through Figure 8, it can be seen that after the short-circuit fault occurs, the terminal voltage of DFIG showed a dynamic process of “first low and then high”. Among them, when a three-phase short-circuit occurred at the inverter bus, the peak overvoltage was the largest, reaching 1.307 p.u., which exceeded the critical value of 1.3 p.u. and may cause DFIG to go off-grid.

In all four simulation scenarios, the terminal voltage of DFIG had been reduced to some extent after adopting the strategy proposed in this paper. And in Case 1 and Case 4, the terminal voltage of DFIG was reduced from 1.307 p.u. and 1.303 p.u. to below 1.3 p.u., respectively, which avoids the off-grid scenario of DFIG and improves the stability of the system.

Through Figure 9 and Figure 10, after the occurrence of a short-circuit fault, it can be seen that during the period of grid voltage drop, the reactive power being supplied by the DFIGs to the grid decreased, the magnitude of the rotor excitation current of the DFIGs increased under the control of the AVR control link, and more reactive power was sent out to the grid to support the terminal voltage; during the period of the grid voltage rise, the magnitude of the rotor excitation current of the DIFGs decreased under the AVR control link, which caused the DFIGs to absorb excess reactive power from the grid to suppress the overvoltage.

However, due to certain lag in the reactive power response of the DFIGs, the droop coefficient of the terminal voltage control link was still the same as in steady-state operation. The magnitude of the rotor excitation current decreased by a small amount and the reactive power absorbed into the grid is small, so the reactive power absorption capacity of the DFIGs cannot be utilized to the maximum extent. On the other hand, the increased reactive power during the low voltage period did not decrease quickly enough, which further aggravated the overvoltage. It can be seen from Figure 9a and Figure 10a that after 15.05 s, VSG-DFIGs were still generating reactive power to drive up the voltage, while the terminal voltage had already risen to 1.1 p.u., indicating a certain lag in reactive power response, and the reactive power generated at this time may have further worsened the transient overvoltage.

Through Figure 9 and Figure 10, it can be seen that regardless of the type and location of the short circuit case, after adopting the strategy proposed in this paper, the output of the rotor excitation current increased during the grid voltage dip, which enabled the VSG-DFIGs to increase the reactive power supply to the grid to meet the grid voltage recovery demand; through Figure 9a during the grid voltage rise, the lag in the reactive power response of the DFIGs was reduced by narrowing the time constant after the DFIG entered a low-voltage fault ride-through, and by switching the droop coefficient of the AVR control link, the DFIGs’ rotor current amplitude was significantly reduced. The most obvious effect was in the case of a three-phase short circuit. The DFIGs absorbed 0.3 p.u. more excess reactive power from the grid than when the strategy proposed was not adopted. Therefore, under the combined effect of time constant reduction and droop coefficient switching, the VSG-DFIGs’ terminal voltage was reduced to 1.26 p.u., which is lower than the threshold value of 1.3 p.u. to trip the system, thus improving the stability of the wind power transmission system via HVDC.

5. Conclusions

In this paper, with the objective of suppressing the transient overvoltage caused by commutation failures in HVDC transmission systems, based on the VSG-DFIG with an AVR control, the time constant and droop coefficient switching strategy of the AVR control link in the abnormal voltage ride-through period is proposed. The effectiveness of the presented strategy was demonstrated through theoretical analysis and simulation verification. Some conclusions are as follows:

(1)

The VSG-DFIG and AVR control link was analyzed, and the influence of the time constant and droop factor on reactive power in the AVR control link was theoretically analyzed: it was found that the amount of reactive power change is directly proportional to the size of the droop coefficient, and that the fundamental reason for transient overvoltage after HVDC commutation faults was the reactive power imbalance in the power grid of the sending end.

(2)

The VSG-DFIG still experienced a delay in the reactive power response, which exacerbated the overvoltage issue and can lead DFIGs disconnecting from grids. When the droop coefficient of the AVR control link remained unchanged, the magnitude of the rotor excitation current decreased by a small magnitude, and the reactive power absorbed from the grid was less, which cannot maximize the exploitation of the DFIG’s reactive power absorption capacity.

(3)

By adjusting the time constant of the AVR control link of the VSG-DFIG, the lag of rotor current and reactive power response can be reduced to a certain extent. By switching the terminal voltage droop coefficients of the AVR control link during the low/high-voltage ride-through periods, the magnitude of the increase or decrease of the VSG-DFIG’s rotor excitation current can be enlarged, so that better reactive power compensation/absorption can be realized, and that the transient overvoltage at the DFIG terminal caused by the commutation failures of HVDC transmission systems can be effectively suppressed.