A rapid detection method for low voltage ride-through in grid-connected inverters

Abstract

This invention discloses a rapid low-voltage ride-through detection method for grid-connected inverters. The method involves acquiring three-phase grid voltages and performing Clarke transform; constructing a decoupled network including positive-sequence and negative-sequence rotating coordinate systems; replacing the traditional low-pass filter with a MAF (Magnetic Frequency Filter) with a window width of half a power frequency cycle in the cross-feedback decoupling path of the positive and negative sequence components to filter out the second harmonic oscillation component under unbalanced operating conditions; calculating the voltage amplitude in real time using the decoupled positive-sequence DC component; and determining the voltage sag fault and its type by comparing the calculated amplitude with a preset threshold. This invention utilizes the notch characteristic of the MAF to completely eliminate second harmonic interference while strictly locking the detection delay within half a power frequency cycle (10ms), significantly improving the response speed and detection accuracy of energy storage inverters during low-voltage ride-through.

Description

Technical Field

[0001] This invention belongs to the field of electrical engineering, and in particular relates to a method for rapid detection of low voltage ride-through in grid-connected inverters. Background Technology

[0002] When the voltage drops in the power grid due to short circuits, lightning strikes, or sudden load changes, the voltage fault can be quickly and accurately identified and compensation control can be triggered to ensure that the inverter maintains grid-connected operation and meets the grid specifications for reactive power support.

[0003] The most commonly used fault detection methods currently include: RMS value detection, three-phase voltage instantaneous value detection, and interference elimination. Axis voltage detection methods include: RMS detection, which judges whether a voltage drop has occurred on the grid side by judging the RMS value within a window (one power frequency cycle) and setting the fault threshold to 85% of the grid voltage rating; and three-phase voltage instantaneous value detection, which directly collects the instantaneous values ​​of the three-phase voltages and determines a voltage drop if the instantaneous voltage of any phase falls below the threshold for 3ms. Interference elimination methods are also included. The shaft voltage detection method starts from a two-phase rotating coordinate system and targets... The shaft voltage detection method cannot function properly under unbalanced grid voltage conditions. Introducing positive and negative sequence components improves reliability. Because phase voltage changes slowly, the input for the first two methods typically converts the three-phase phase voltages to three-phase line voltages to improve response speed.

[0004] The patent with publication number CN112448412A proposes a method based on A sliding window filtering method for low-voltage ride-through determination of the DC component of the axial line, which uses real-time sampled grid-side voltage... After decomposition, sliding window filtering is performed, with the window length being the number of sampling points within the power grid cycle. The method calculates the average voltage within the sliding window and compares it with a set threshold. If the average voltage is less than the threshold, a low-voltage ride-through is considered to have occurred. While this method can reliably determine the drop depth, it heavily relies on mean filtering, and the separation of positive and negative sequences introduces a delay of at least 1 / 4 of the power frequency cycle. Therefore, the response speed of this method is limited and cannot meet the requirements for rapid detection.

[0005] Patent CN102854421A discloses a rapid low-voltage ride-through detection scheme for photovoltaic inverters, which obtains the instantaneous values ​​of the three-phase grid voltage through sampling. Voltage is determined by Calculations were performed, and a voltage sag threshold was set. Minimum withstand voltage limit By judgment The system determines whether the voltage is below a certain threshold and takes appropriate action accordingly. Hardware-wise, it uses an AD conversion unit based on a 2812 digital processing chip to shorten the single-phase voltage drop delay detection time to within 5ms. However, this strategy lacks robustness under shallow voltage drops and is prone to misjudgment when the grid voltage contains harmonics or flicker occurs.

[0006] Patent CN104237711A addresses the issue of misjudgment caused by shallow drops in instantaneous values. This invention uses a two-phase rotating coordinate system to extract the positive and negative sequence components of the acquired three-phase voltage instantaneous values. The expression is as follows:

[0007] In the formula, , These represent the positive and negative sequence of the grid-side voltage, respectively. Axial components, The value is a constant. The fault is detected by judging that the extracted positive sequence component is less than the set grid voltage drop threshold. Compared with the previous method, this method improves the accuracy of judging grid voltage harmonics, flicker and shallow drops, and the judgment time is further shortened. However, during the voltage recovery stage, this method is prone to false triggering. Summary of the Invention

[0008] To address the aforementioned problems, this invention discloses a rapid low-voltage ride-through detection method for grid-connected inverters.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows: A method for rapid low-voltage ride-through detection of a grid-connected inverter includes the following steps: Step 1: Collect the three-phase instantaneous voltage signals of phases A, B, and C of the power grid. , , And transform it using an equal amplitude transformation matrix to Two-phase stationary coordinate system shaft and shaft voltage ; Indicates time; Step 2: Construct a decoupling network, the decoupling network including angular velocity... Clockwise rotating positive sequence synchronous rotating coordinate system and angular velocity Counterclockwise rotating negative sequence synchronous rotating coordinate system ; Step 3: Use the Park transformation matrix to... Projected onto the positive sequence synchronous rotating coordinate system respectively Synchronous Rotation Coordinate System with Negative Sequence In the process, the positive sequence voltage estimate after oscillation elimination is calculated in real time. and negative sequence voltage estimates ; Step 4: Apply a moving average filter (MAF) to... and Filtering is performed to obtain a pure positive-sequence DC component. and pure negative sequence DC component ;in, Positive sequence without second harmonic components DC component of the axis Positive sequence without second harmonic components DC component of the axis The negative sequence does not contain a second harmonic component. DC component of the axis The negative sequence does not contain a second harmonic component. The DC component of the axial direction; where the size of the sliding window during filtering is half the period of the fundamental frequency of the power grid; Step 5, according to , and , Calculate the positive sequence voltage amplitude and negative sequence voltage amplitude ; Step Six, if If the duration exceeds the anti-jitter time, a voltage dip fault is determined to have occurred in the power grid, and an LVRT start signal is generated; for A1 and A2 type energy storage converters, It is 90% of the rated voltage; for B1 and B2 type energy storage converters, 85% of the rated voltage; Step 7: After generating the LVRT start signal, if The fault was determined to be a symmetrical drop fault, and the control system entered a symmetrical reactive power support mode; if The fault was identified as an asymmetric drop fault, and the control system entered the unbalanced compensation mode. The preset imbalance threshold is obtained from actual experimental data.

[0010] A further improvement is made in step one, where the Clarke transformation is used to transform the three-phase stationary coordinate system. Converting the voltage to a two-phase stationary coordinate system The voltage vector in the stationary coordinate system below is denoted as , The method to obtain it is as follows: .

[0011] In a further improvement, in step three, the coordinate system is rotated synchronously in the forward sequence. The projection below is obtained through the rotation factor. Perform the transformation; where, The imaginary unit, Let t be the fundamental angular frequency of the power grid, and t represent time. The electric angle represents the change over time; the orthogonal projection after transformation by the rotation factor is as follows:

[0012] Expanded into matrix form, the real part corresponds to Axis, imaginary part corresponding axis:

[0013] in, , The first term is the extracted pure positive-sequence DC component, while the second term is derived from the negative-sequence voltage. and The frequency caused is Interference items; Negative sequence synchronous rotating coordinate system Projection over rotation factor Perform the transformation:

[0014] Expand into matrix form:

[0015] in, , It is pure. shaft and The second term is derived from the negative sequence DC component, and the third term is derived from the positive sequence voltage. and The frequency caused is Interference items;

[0016] in, Positive sequence of grid voltage Axial components and positive sequence Two-dimensional vectors with axial components. The negative sequence of the grid voltage Axial components and negative sequence A two-dimensional vector with axial components; For pure negative order DC component of axis and A two-dimensional vector of the DC component of the axis. For pure or orthogonal order DC component of axis and A two-dimensional vector of the DC component of the axis; This is the positive-sequence voltage estimation vector obtained through the decoupling network. This is the negative sequence voltage estimation vector obtained through the decoupling network; definition The second harmonic rotation transformation matrix is:

[0017] for The transpose of .

[0018] Further improvements, in step four, involve adjusting the moving average filter (MAF)... and Sliding window during filtering The size is: Sliding window, The fundamental frequency period of the power grid; The time-domain discretization recursive equation for the moving average filter (MAF) is:

[0019] in The number of sampling points, i.e. , Sampling frequency, Indicates the current time raw input Voltage sample value, This indicates the result after passing through the moving average filter (MAF). Voltage output.

[0020] A further improvement is made in step five, where the positive sequence voltage amplitude... and negative sequence voltage amplitude The calculation method is as follows:

[0021] .

[0022] Advantages of this invention: 1. This invention proposes a simple and effective detection method to determine whether a voltage drop fault has occurred at the grid connection point.

[0023] 2. The MAF proposed in this invention is a finite impulse response (FIR) filter, and its settling time is strictly equal to the window width (10ms). Compared with the asymptotic convergence of the LPF, this method can guarantee an error-free detection result within 10ms after the fault occurs.

[0024] 3. The novel filter proposed in this invention has infinite attenuation gain for the second harmonic component, which solves the problem that traditional LPF may still have residual ripple even in steady state, and ensures the detection accuracy under asymmetric faults.

[0025] 4. The MAF implemented in this invention using a recursive approach has a very small computational load, making it very suitable for running in low-cost DSPs or MCUs. Attached Figure Description

[0026] Figure 1 Comparative waveforms for fault detection methods when three-phase voltage symmetrically drops to 72V.

[0027] Figure 2 The waveforms are comparisons of fault detection methods when the voltage of phase a drops to 72V.

[0028] Figure 3 The fault waveform is the fault quantity waveform of the fault detection method when the voltage of phase a drops to 72V.

[0029] Figure 4a This is an overall waveform diagram of the experiment when the three-phase voltage drops symmetrically to 60V.

[0030] Figure 4b This is a detailed experimental waveform diagram when the three-phase voltage drops symmetrically to 60V.

[0031] Figure 5a This is an overall waveform diagram of the experiment when the voltage of phase b and phase c drops to 120V.

[0032] Figure 5b Detailed experimental waveforms of phase b and phase c voltages dropping to 120V.

[0033] Figure 6a This is the overall waveform diagram of the experiment when the voltage of phase a drops to 60V.

[0034] Figure 6b This is a detailed experimental waveform diagram when the voltage of phase a drops to 60V. Detailed Implementation

[0035] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0036] Example The present invention is as follows: Since the standard has requirements for the response time of reactive current, the fault detection system should be able to detect the occurrence of the fault as quickly as possible after a fault occurs. In addition to meeting the requirements of speed, it should also be able to respond normally to faults with shallow depths.

[0037] First, the three-phase instantaneous voltage signals of the power grid are acquired using voltage sensors. , , The Clarke transformation is used to transform the three-phase stationary coordinate system. Converting the voltage to a two-phase stationary coordinate system The voltage vector in the stationary coordinate system below is denoted as .

[0038] To ensure that the magnitude of the transformed voltage vector is equal to the amplitude of the phase voltage, this embodiment uses an equal-amplitude transformation matrix:

[0039] in,

[0040] It should be noted that the above Given the total stationary coordinate voltage vector obtained through Clarke transformation, the following formula does not recalculate this voltage vector, but rather decomposes it into positive and negative sequence components. In the case of grid voltage imbalance (i.e., the presence of a negative sequence component), the above stationary coordinate voltage vector... It can be further decomposed into positive-order components. With negative order components The superposition of, that is: +

[0041] This can be further written as:

[0042] in, , These represent the positive and negative sequence voltage amplitudes, respectively. , This is the initial phase.

[0043] This embodiment constructs two synchronous coordinate systems rotating in opposite directions to separate the positive and negative sequence components, wherein the positive sequence synchronous rotating coordinate system ( ): In terms of angular velocity Rotate counterclockwise, negative sequence synchronous rotation coordinate system ( ): In terms of angular velocity Rotate counterclockwise.

[0044] Using the Park transformation matrix to Projected onto these two coordinate systems respectively: Projection in positive order coordinate system ( ) via rotation factor Perform the transformation:

[0045] Expand into matrix form (real parts correspond) Axis, imaginary part corresponding axis):

[0046] in, , The first term is the extracted pure positive-sequence DC component, while the second term is derived from the negative-sequence voltage. and The frequency caused is Interference at 100Hz.

[0047] Negative sequence synchronous rotating coordinate system Projection over rotation factor Perform the transformation:

[0048] Expand into matrix form:

[0049] in, , It is pure. shaft and The second term is derived from the negative sequence DC component, and the third term is derived from the positive sequence voltage. and The frequency caused is Interference term (100Hz); definition The second harmonic rotation transformation matrix is:

[0050] Its transpose matrix is .

[0051] Real-time calculation of positive sequence voltage estimates after oscillation elimination using a decoupling network and negative sequence voltage estimates :

[0052] in, and These are instantaneous sampled values ​​containing noise; and This is the purified tributary feedback value after MAF filtering. To completely filter out... (100Hz) component and achieve the fastest response for the decoupled signal and Using half the fundamental frequency period of the power grid, i.e. A sliding window is used for filtering. The time-domain discretization recursive equation for MAF is:

[0053] in The number of sampling points, i.e. , Sampling frequency, Indicates the current time raw input Voltage sample value, This indicates the result after passing through the moving average filter (MAF). Voltage output.

[0054] The oscillation frequency caused by the imbalance is 100Hz, and its period is exactly 10ms. By performing a moving average within a 10ms window, the integral value of this oscillation component is always zero. Therefore, the MAF output... It is a completely ripple-free DC component, and the transient response time of the filtering process is strictly fixed at 10ms.

[0055] After MAF filtering, a clean positive-sequence component is obtained. , and negative order components , Therefore, the positive-sequence voltage amplitude and the negative-sequence voltage amplitude can be calculated as follows:

[0056]

[0057] When a symmetrical voltage drop occurs in the power grid, detection Is it less than the drop threshold required by national standards? .like If the duration exceeds the anti-jitter time, a voltage dip fault is determined to have occurred in the power grid, and an LVRT start signal is generated. After generating the LVRT start signal, further comparisons are performed. Compared with the preset imbalance threshold The size. If The fault was determined to be a symmetrical drop fault, and the control system entered a symmetrical reactive power support mode; if The fault was identified as an asymmetrical drop fault, and the control system entered the unbalanced compensation mode.

[0058] The performance of the MAF-based DDSRF separation detection method for RMS and instantaneous value detection is compared and analyzed based on simulations under two operating conditions: symmetrical voltage drop and asymmetrical voltage drop. Figure 1 , Figure 2The simulation comparison waveforms of fault flag bits obtained by the RMS detection method, the instantaneous three-phase voltage detection method, and the MAF-based DDSRF separation detection method are shown for three-phase voltage drops from 220V to 72V and phase A voltage drops from 220V to 72V, respectively. In the figures, rms, u_d, and ddsrf represent the values ​​obtained by the three-phase voltage instantaneous value detection method, the interference removal method, and the other method, respectively. Fault flags obtained by shaft voltage detection method and MAF-based DDSRF separation detection method.

[0059] according to Figure 1 It can be seen that the three-phase voltage instantaneous value detection method has the fastest detection speed, followed by the MAF-based DDSRF separation detection method, and lastly, the interference elimination method. Shaft voltage detection method. Eliminating interference. The shaft voltage detection method exhibits more false triggering during faults because its input is obtained from the positive-negative sequence separation stage and does not filter out the second harmonic component. At the instant of voltage change, the obtained positive-negative sequence components show overshoot and only reach steady state after a certain period. Therefore, the fault quantity changes significantly during the adjustment process, causing false triggering in the fault detection stage. The other two methods did not exhibit false triggering under symmetrical voltage dips. Therefore, the three-phase voltage instantaneous value detection method performs best when there is a symmetrical voltage dip on the grid side.

[0060] according to Figure 2 It can be seen that under the condition of voltage asymmetry drop, the DDSRF separation detection method based on MAF has the fastest detection speed, followed by the interference elimination method. The first two methods, shaft voltage detection and the last one, are instantaneous three-phase voltage detection. The detection speeds of the first two methods are basically the same. However, from... Figure 3 As can be seen, when the grid-side voltage experiences an asymmetrical drop, the fault quantity obtained by the three-phase voltage instantaneous value detection method will contain a second harmonic component. Since the fault threshold is DC, this leads to false triggering of the three-phase voltage instantaneous value detection method, affecting the judgment of grid-side voltage recovery. Therefore, when an asymmetrical voltage drop occurs on the grid side, the MAF-based DDSRF separation detection method exhibits the best detection speed performance.

[0061] Based on the above analysis, the MAF-based DDSRF separation detection method is finally adopted as the fault detection method in low voltage ride-through.

[0062] According to the detection method of this invention, the experimental waveforms of the three-phase voltage dropping from 220V to 60V, the voltage of phases B and C dropping from 220V to 120V, and the voltage of phase A dropping from 220V to 60V in a 330kW energy storage converter are as follows: Figures 4a-6bAs shown, the strategy of this invention can achieve stable and accurate fault diagnosis under both symmetrical and asymmetrical voltage drops on the grid side, with a rapid response throughout the process and the avoidance of false triggering.

[0063] This invention sets the integration window length of MAF to be (That is, 1 / 2 of the 50Hz power grid cycle). The interference component caused by power grid imbalance has a frequency of 100Hz (period of 10ms), and the average value of this interference component is always zero within the 10ms window. Utilizing this characteristic, MAF exhibits ideal "notch filter" characteristics, which can instantly (after the end of the window) completely eliminate interference and obtain a pure DC component.

[0064] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. Other modifications can be easily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A method for rapid low-voltage ride-through detection of a grid-connected inverter, characterized in that, Includes the following steps: Step 1: Collect the three-phase instantaneous voltage signals of phases A, B, and C of the power grid. , , And transform it using an equal amplitude transformation matrix to Two-phase stationary coordinate system shaft and shaft voltage ; Indicates time; Step 2: Construct a decoupling network, the decoupling network including angular velocity... Clockwise rotating positive sequence synchronous rotating coordinate system and angular velocity Counterclockwise rotating negative sequence synchronous rotating coordinate system ; Step 3: Use the Park transformation matrix to... Projected onto the positive sequence synchronous rotating coordinate system respectively Synchronous Rotation Coordinate System with Negative Sequence In the process, the positive sequence voltage estimate after oscillation elimination is calculated in real time. and negative sequence voltage estimates ; Step 4: Apply a moving average filter (MAF) to... and Filtering is performed to obtain a pure positive-sequence DC component. and pure negative sequence DC component ;in, Positive sequence without second harmonic components DC component of the axis Positive sequence without second harmonic components DC component of the axis The negative sequence does not contain a second harmonic component. DC component of the axis The negative sequence does not contain a second harmonic component. The DC component of the axial direction; where the size of the sliding window during filtering is half the period of the fundamental frequency of the power grid; Step 5, according to , and , Calculate the positive sequence voltage amplitude and negative sequence voltage amplitude ; Step Six, if If the duration exceeds the anti-jitter time, a voltage dip fault is determined to have occurred in the power grid, and an LVRT start signal is generated; for A1 and A2 type energy storage converters, It is 90% of the rated voltage; for B1 and B2 type energy storage converters, 85% of the rated voltage; Step 7: After generating the LVRT start signal, if The fault was determined to be a symmetrical drop fault, and the control system entered a symmetrical reactive power support mode; if The fault was identified as an asymmetric drop fault, and the control system entered the unbalanced compensation mode. The preset imbalance threshold is obtained from actual experimental data.

2. The method for rapid low-voltage ride-through detection of a grid-connected inverter as described in claim 1, characterized in that, In step one, the Clarke transformation is used to transform the three-phase stationary coordinate system. Converting the voltage to a two-phase stationary coordinate system The voltage vector in the stationary coordinate system below is denoted as , The method to obtain it is as follows: 。 3. The method for rapid low-voltage ride-through detection of a grid-connected inverter as described in claim 1, characterized in that, In step three, the coordinate system rotates synchronously in the forward sequence. The projection below is obtained through the rotation factor. Perform the transformation; where, The imaginary unit, Let t be the fundamental angular frequency of the power grid, and t represent time. The electric angle represents the change over time; the orthogonal projection after transformation by the rotation factor is as follows: ; Expanded into matrix form, the real part corresponds to Axis, imaginary part corresponding axis: ; in, , The first term is the extracted pure positive-sequence DC component, while the second term is derived from the negative-sequence voltage. and The frequency caused is Interference items; Negative sequence synchronous rotating coordinate system Projection over rotation factor Perform the transformation: ; Expand into matrix form: ; in, , It is pure. shaft and The negative sequence DC component, the second term is composed of the positive sequence voltage. and The frequency caused is Interference items; ; in, Positive sequence of grid voltage Axial components and positive sequence Two-dimensional vectors with axial components. The negative sequence of the grid voltage Axial components and negative sequence A two-dimensional vector with axial components; For pure negative order DC component of axis and A two-dimensional vector of the DC component of the axis. For pure or orthogonal order DC component of axis and A two-dimensional vector of the DC component of the axis; This is the positive-sequence voltage estimation vector obtained through the decoupling network. This is the negative sequence voltage estimation vector obtained through the decoupling network; definition The second harmonic rotation transformation matrix is: ; for The transpose of .

4. The method for rapid low-voltage ride-through detection of a grid-connected inverter as described in claim 1, characterized in that, In step four, the moving average filter (MAF) is applied. and Sliding window during filtering The size is: Sliding window, The fundamental frequency period of the power grid; The time-domain discretization recursive equation for the moving average filter (MAF) is: ; in The number of sampling points, i.e. , Sampling frequency, Indicates the current time raw input Voltage sample value, This indicates the result after passing through the moving average filter (MAF). Voltage output.

5. The method for rapid low-voltage ride-through detection of a grid-connected inverter as described in claim 1, characterized in that, In step five, the positive sequence voltage amplitude and negative sequence voltage amplitude The calculation method is as follows: ; 。

Images