Protection for AC transmission line between DFIG wind farm and MMC station

The characteristics of power electronic controlled sources are reflected at both ends of the AC line applied to the integration of long-distance wind farms into the MMC-HVDC converter station, and the fault characteristics of the system have undergone fundamental changes. For the doubly-fed wind power AC transmission line with a modular multilevel converter, the short-circuit current characteristics provided by the power supply at both ends of the line lead to the inadaptability of longitudinal differential protection. In view of the significant difference in the complexity of the short-circuit current component provided by the doubly-fed wind power and the converter at both ends of the AC transmission line, a line pilot protection idea based on the complexity of the current component is proposed. The complexity of the current component is characterized by singular entropy theory, and then the singular entropy algorithm and its scheme for pilot protection are constructed. A refined electromagnetic transient model of MMC-HVDC for doubly-fed wind farms is built in PSCAD for simulation verification. The simulation results show that the proposed method can quickly and reliably identify various types of internal and external faults, and solve the problem of incorrect action of traditional protection in this scenario. In addition, the scheme is not affected by fault type, location, or transition resistance.


INTRODUCTION
With the widespread global support for the 'Paris Agreement', renewable energy has accounted for about 60% of global new energy generation in the past five years.Many countries and regions around the world have proposed carbon neutrality goals to promote the development and utilization of clean and low-carbon energy [1][2].China is rich in wind power resources, boasting the advantages of being suitable for large-scale development, high utilization hours, and being close to the load center.Its development and utilization are an important part of China's response to climate change and the promotion of energy transformation and upgrading.Compared with conventional DC transmission, MMC-HVDC has the advantages of flexible control of active and reactive power, easy formation of DC grid capacity, low switching loss, high fault tolerance, strong fault ride-through capability, black start capability and no commutation failure risk, so it is more suitable for wind power access.However, the addition of wind power and MMC-HVDC also brings challenges to traditional relay protection.At present, the research on AC side protection of wind power flexible grid-connected systems mainly focuses on the flexible grid-connected end.One end of the system is a power electronic power supply (flexible DC converter), and the other end is a synchronous generator.Since the synchronous power generation system can provide strong voltage support, after the system failure, the flexible DC enters the low voltage ride through control which can provide stable fault current to the AC system [5][6][7][8].Therefore, experts and scholars put forward the new protection principle based on fault state signal and the traditional protection setting principle [9][10], the new protection principle based on single terminal adaptive, and the new differential protection principle.However, the basis of these new principles counts on one side of the system has a synchronous power supply to provide voltage support and a large short-circuit current.However, the basis of these new principles depending on one side of the system has a synchronous power supply to provide voltage support and a large short-circuit current.In the scenario where the converter cannot provide stable fault current, the proposed protection principles have the problems of performance degradation and even failure to correctly identify faults.Relatively speaking, the research on line protection of power electronic power supply on both sides is not yet mature.It is urgent to study the AC line protection principle of wind power integration into a flexible DC converter station.Therefore, the study of double-fed wind power AC line protection schemes with MMC converter stations still has important research significance.This paper investigates the pilot protection of AC transmission lines with doubly-fed wind power and MMC converter stations.First, the transient features of short-circuit current from doubly-fed wind power and MMC converter stations are analyzed.Then, based on complexity, the singular entropy algorithm is applied to extract the characteristic difference of fault current on both sides of the line and design the pilot protection scheme.

DFIG fault characteristics
A wound-rotor induction motor that connects its stator winding directly to the grid and its rotor winding to the grid via a back-to-back converter is called a Doubly Fed Induction Generator (DFIG).The converter comprises two independent control parts and is interconnected by DC side capacitors.The structure is shown in Figure 1.According to the different severity of the fault, the doubly-fed wind turbine has two different low voltage ride-through modes: Crowbar circuit input and transient control strategy, so there are two different fault current characteristics.When a serious fault occurs in the doubly-fed AC transmission line of the associated MMC converter station, the current on the line increases instantaneously, and the stator voltage of the wind turbine will drop to a certain extent.The rotor is connected in series with the Crowbar circuit, and the rotor-side converter is short-circuited.The rotor voltage drops to 0. The DFIG short-circuit current expression is as follows: where A, B, and D are defined as the amplitudes of the three components in the current expression.
When the fault of the sending line is not very serious, that is, the rotor-side short-circuit current caused by the fault is small, the corresponding DFIG terminal voltage drop is also light, and the change of the short-circuit current and voltage cannot reach the threshold of the Crowbar circuit being activated.At this time, the Crowbar circuit will not start, and the DFIG will change from normal excitation control to fault ride-through control mode.The power factor is changed by converter control, the active power output is reduced, and the reactive power output is increased to meet the requirements of lowvoltage ride-through.At this time, the short-circuit current expression is: where A 2 , B 2 , C 2 , D 2 , E 2 and F 2 are the amplitudes of the corresponding components; k 1 and k 2 are different parameters depending on the control method.
According to the short-circuit current expressions (1) and ( 2), it can be seen that the short-circuit current has many components and strong component complexity.It is composed of kinds of currents, which include attenuated DC, sinusoidal AC, and attenuated AC, and the amplitude of each component changes with different control strategies.Short-circuit current frequency will be affected by three components, no longer power frequency; the current amplitude and phase angle under different conditions are also affected by the three components, with uncertainty.In addition, different fault ridethrough control strategies will lead to different characteristics, and it is difficult to find the common points of fault characteristics through analysis.

MMC-HVDC converter fault characteristics
Since the system side MMC converter station and the doubly fed wind power AC transmission line are isolated by the DC line and the wind farm side MMC converter station, the distance is far.When the double-fed wind power AC transmission line fails, the system side MMC converter station has little effect on the short-circuit current fault characteristics of the wind power AC transmission line.Therefore, the influence of the control strategy of the system side MMC converter station is ignored, and only the influence of the wind farm side converter station on the short-circuit current fault characteristics of the wind power AC line is taken into consideration.And wind farm side converter adopts an AC voltage frequency control method to provide voltage support for the wind farm.After an AC side fault, to prevent the converter from overcurrent, a low-voltage current limiting control strategy will be adopted, that is, using voltage control, but reducing the current-voltage reference value.After adjustment by a PI regulator, the output current will be reduced.Therefore, after a fault, the MMC converter presents a voltage source characteristic.Since the converter transformer isolates the zerosequence current, the short circuit current provided by the flexible direct converter is: Equation (4) represents the analytical formula for asymmetric faults; Equation (5) represents an analytical expression for symmetric faults; Emmc (t) is the internal potential of the flexible DC converter; Z eq is the equivalent impedance related to the fault type.For single-phase faults, Z eq =Z + +Z - +Z 0 , for two-phase interphase faults, Z eq =Z + +Z -, and for two-phase ground faults, Z eq =Z + +(Z -×Z 0 )/ (Z - +Z 0 ), where Z + , Z -, and Z 0 are positive sequence equivalent impedance, negative sequence equivalent impedance, and zero sequence equivalent impedance, respectively; ω + and ω -are the fundamental frequency positive sequence angular frequency and the fundamental frequency negative sequence angular frequency.Theoretically, it can be seen from Eqs. ( 3) and ( 4) that the short-circuit current output by the MMC converter presents a sinusoidal principal component, and the waveform complexity is weak.
According to the above-mentioned wind power side fault current expression and the fault current expression of the flexible wind field side, it can be seen that when a fault occurs, the current components provided by the wind power side are more complex, including many components that decay with time, and the waveform changes with time.It is irregular and has a large gap with the steady-state sinusoidal waveform, showing the characteristics of weak feed and uncertain phase.The composition of the short-circuit current provided by the MMC side is relatively simple, and the waveform changes regularly with time.The gap with the steady-state sine waveform is small, showing weak feed and phase uncertainty.

SINGULAR ENTROPY CHARACTERIZATION OF CURRENT COMPLEXITY
The above analysis shows that both ends of the AC transmission line with doubly-fed wind power and MMC converter station have power electronic converter devices, but the structure and principle of the power supply at each end are different.The short-circuit current from the doubly-fed wind power and the short-circuit current from the MMC converter station are very different and uncertain when a line fault occurs.Owing to the unpredictability of the short-circuit current supplied by power sources at each end, using current differential protection on the line may result in rejection.While the short-circuit current supplied by both power sources varies based on the fault condition or control strategy, the discrepancy between the current components at each end remains unaffected by these factors.The complexity of the short-circuit current does not change with alterations in the fault condition or control strategy either.The short-circuit current supplied by the wind power side differs significantly from that supplied by the flexible side; however, no precise method is available for extracting the current components.As the components of the short-circuit current on each side differ, the information they hold also varies.Therefore, the distinction between short-circuit currents on both sides can be characterized by examining the complexity of their components.
Complexity is often used to objectively measure the complexity of the measured electronic signal or system and reflects the nonlinear characteristics of the electronic signal or system.Singular entropy can quantitatively represent the structural complexity of an instantaneous sequence through the lens of complexity analysis.This approach is better suited for extracting signal features and describing system characteristics, without demanding a high data sampling rate.Singular entropy has found extensive applications in extracting complexity from mechanical vibration signals.In the absence of a line fault or external fault, the current flowing through both ends of the line consists of the same steady-state power frequency current, leading to identical component complexity at each end.Consequently, the discrepancy in current complexity at both ends of the line can be employed to develop pilot protection.
The sampled data from each phase current at both ends of the transmission line can create a onedimensional signal.For instance, considering the current sampling signal of any phase on the wind farm side, the current sampling values over a specific time period can be organized into a one-dimensional signal, denoted as.Using the singular spectrum entropy analysis of one-dimensional time series can objectively measure the complexity of the measured signal or system and reflect the nonlinear characteristics of the signal or system.This paper proposes the principle of longitudinal differential protection by using singular spectrum entropy.
First, it needs to construct a Pattern Matrix: According to the time series composed of current signals, the mode or trajectory matrix is constructed, and the analysis window length M and time delay constant r are selected.The sample data is intercepted in the window order of (M, r), that is, the order of I=ik is divided into c segments, and the pattern matrix A is formed in turn: The sampling frequency of 5 kHz is selected, that is, within a time window, the number of sampling points N is 100.We have 100 discrete current signals, embedding dimension M being 5; with the delay constant r being 1, we set the matrix A as a matrix of m rows and M columns.
The current signals of each phase on both sides of the line can be written in the form of a onedimensional time series, and then the singular value decomposition of the current mode matrix is performed, as shown in Equation ( 6): The diagonal elements of S are arranged in the order from small to large to obtain a one-dimensional array, which is taken into the formula to obtain the singular entropy of the current signal.The size of the array element represents the proportion of the patterns it represents in the whole.If let: Then p i is the proportion of the i-th singular value in the whole singular value spectrum.Also, p i represents the proportion of the i-th mode of matrix A in all modes.Therefore, according to the definition of information entropy, the singular entropy of the DFIG side and the MMC side in the time domain can be obtained as follow: where H DFIG is the Singular Entropy Value of the Current at the Installation of the First Circuit Breaker on the DFIG Side; H MMC is the singular entropy value of the current at the installation of the MMC side end circuit breaker.And the singular entropy calculation is performed after current sampling at both ends of the studied line.Next, the singular entropy of the short-circuit current supplied by the doublyfed wind power is deducted from the singular entropy of the short-circuit current supplied by the MMC.The computed results can be observed in Table 1 1, k represents the multiple relationships between the singular entropy difference of internal fault and the singular entropy difference of external fault.As observed in Table 1, the magnitude of the difference in current singular entropy at both ends of the line is at three decimal places during normal operation or external fault.However, when an internal fault arises, the difference in magnitude of current singular entropy at both ends of the line is at one decimal place.Consequently, the difference in singular entropy can reliably discern between internal and external faults under various fault conditions.
To sum up, the pilot protection criterion of doubly fed wind power AC transmission line of the associated MMC converter station can be constructed: To facilitate observation and analysis, the difference is multiplied by a coefficient of 100 to obtain.
Hset is the threshold value.Considering that the maximum transmission error of the current sensor is 10% and a certain margin is left, that value must be larger than the maximum value when the line has an external fault.At the same time, it must be smaller than the minimum value when the transmission line has an internal fault, so as to ensure that the protection device does not refuse to operate and does not malfunction after 15 simulation tests.

SIMULATION TEST AND RESULT ANALYSIS
In order to verify the correctness of the proposed scheme, the simulation model of the doubly-fed wind power AC transmission system of the connected MMC converter station is built in PSCAD.This section simulates and analyzes the operation performance of the protection proposed in this paper when different types of faults occur at different locations of the doubly-fed wind power AC transmission line of the associated MMC converter station.The simulation structure is shown in Figure 2.For the selection of simulation sampling rate, the sampling rate of traditional microcomputer protection devices is generally 1 kHz and 1.2 kHz.The sampling rate of protection devices in smart substations is high, at 4 kHz and 5 kHz.In the application scenario of this paper, 1.2 kHz can meet the calculation requirements of the algorithm while ensuring the economy.Therefore, 1.2 kHz is selected as the sampling rate of the simulation model.For the choice of the time window, this article selects a 20 ms time window.

Different fault types
Three-phase short circuits, AB phase-to-phase and AB grounding, and single-phase grounding are set at the midpoint of the line.The fault occurs at 2 s, which verifies the effect of low-frequency differential momentum under different types of faults.The results are shown in Figure 3.
According to Figure 3, the proposed main protection criterion can be started accurately at 20 ms in various types of fault scenarios, which avoids the problem of incorrect operation of traditional protection due to excessive current phase angle difference at both ends.

Different fault locations
The sampled current at both ends of the line changes with the fault location, which influences the performance of the protection scheme.This paper tests the proposed scheme for three-phase shortcircuit faults at different line positions.Figure 4 shows the simulation results.From Figure 4, it can be seen that when a fault occurs at any position of the line, regardless of whether the fan adopts various low-voltage ride-through methods, the protection action value is far greater than the protection criterion, and the proposed protection scheme can operate reliably and quickly.

Different fault resistance
The most common fault in power systems is a ground fault.When a ground fault occurs, there is often a transition resistance.From the perspective of the safe operation of power systems, the protection principle must correctly identify faults when non-metallic faults occur.The fault types are set to singlephase grounding and two-phase grounding at the first end of the line, and the transition resistances are set to 50,100, and 150 Ω respectively.The singular spectrum entropy calculation results of the fault occurring in 2s~2.02s are shown in Table 2.
With the increase of the transition resistance, the change degree of the current signal is different, and the singular spectrum entropy will also increase.Since the maximum transition resistance of the 220 kV system is 100 Ω, the proposed protection still has good performance in high resistance fault scenarios.

Effect of noise
The power system always has random noise, and the protection algorithm needs to be able to resist noise.It is generally believed that the maximum signal-to-noise ratio of noise is 25 dB, so the noise scenarios of 35, 25, and 15 dB are selected to verify the proposed protection principle.As shown in Table 3, the fault phase is not affected much by the noise, and the protection can accurately detect the fault.For the non-fault phase, the noise will cause the current difference on both sides, which results in the increase of singular spectrum entropy.However, in the 25dB high noise scene, the protection can still correctly distinguish the fault type, which can prove that the proposed method has a good anti-noise ability.

Conclusion
Due to the unstable current phase angle difference after the fault of the doubly-fed wind power AC line of the connected MMC converter station with power electronic converter equipment on both sides, the traditional protection does not operate correctly.Aiming at this problem, this paper proposes a pilot protection principle based on singular entropy, and draws the following conclusions: 1) Compared with traditional differential protection, the proposed protection method can correctly identify faults when the current angles at both ends of the line are quite different.By utilizing the advantage that the singular value spectrum entropy can extract fault features without relying on the current phase angle, the problem of incorrect action of traditional protection due to the large phase angle difference of fault current is solved.
2) The proposed method can reliably and correctly identify various types of faults inside and outside the region and still has good recognition ability under high resistance faults and noise.

Figure 3 .
Figure 3. Dynamic performance of protection under different fault types scenarios

Figure 4 .
Figure 4. Influence of Different Fault Locations on Protection Scheme

Table 1 .
. Fault current singular entropy of different fault types

Table 2 .
Protection performance under different fault resistance

Table 3 .
Protection performance under different noise