A Microgrid Cooperative Control Method Based on Power Oscillation Mode Identification
By performing power oscillation mode identification in the microgrid, identifying the return branch pairs between parallel branches and generating active and reactive power pair corrections, the problem of identifying and suppressing power oscillations in the parallel operation of multiple converters is solved, and effective power oscillation control is achieved.
Patent Information
- Application Number
- CN202611442856.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-17
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-17
AI Technical Summary
In a microgrid with multiple converters operating in parallel, low-frequency power oscillations can propagate along branches and the common AC bus, making it difficult for the controller to determine which branch pairs mainly bear the round-trip power. Vibration suppression correction may be dispersed to non-main branches, and existing control methods cannot effectively suppress power oscillations without increasing the types of sensors or rewiring.
By synchronously collecting the active and reactive power of the parallel branches, as well as the bus voltage and frequency, power oscillation mode identification is performed to determine the dominant frequency and branch response point. Return branch pairs are identified, and active and reactive power pair corrections are generated. Amplitude limiting is then performed to suppress power oscillations.
It effectively identifies the round-trip power paths between parallel branches, reduces branch power fluctuations and current stress, and keeps the total active and reactive power correction of the microgrid constant, avoiding the need for additional sensors and line modifications.
Smart Images

Figure CN122951263B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of microgrid control technology, and more specifically, to a microgrid collaborative control method based on power oscillation mode identification. Background Technology
[0002] AC microgrids typically connect distributed generation converters, energy storage converters, and loads, with each converter connected in parallel to a common AC bus via AC branches. During operation, primary control handles rapid voltage and current regulation, while secondary collaborative control handles frequency and voltage restoration, as well as active and reactive power sharing. Control references are usually generated from local quantities, adjacent quantities, common bus quantities, or total network quantities.
[0003] When multiple converters operate in parallel, low-frequency power swings propagate along the branches and the common AC bus. If two parallel branches output in opposite directions within the same swing frequency band, the common AC bus measurement may show a small fluctuation, but there is still round-trip power between the two branches, and the RMS value of the branch current, power swing, and capacity margin will still be affected. Relying solely on the combined or adjacent average values of the common AC bus makes it difficult for the controller to determine which branch pair mainly bears the round-trip power, and vibration suppression correction may also be distributed to non-primary branches.
[0004] In terms of engineering, the new vibration suppression treatment also needs to be compatible with the existing droop primary control, secondary frequency and voltage regulation, capacity limiting and communication cycle, without increasing the types of sensors, without changing the wiring and bus, and without changing the total active and total reactive power correction targets of the microgrid. Summary of the Invention
[0005] This invention provides a microgrid collaborative control method based on power oscillation mode identification, which solves the technical problems mentioned in the background art.
[0006] A microgrid collaborative control method based on power oscillation mode identification, applied to a microgrid with multiple converters connected in parallel to a common AC bus, includes:
[0007] The active and reactive power of each parallel branch, as well as the bus voltage and bus frequency, are collected synchronously to form a data window;
[0008] Power oscillation mode identification is performed on the active power of the branch to determine the dominant power oscillation mode and dominant frequency;
[0009] Active power oscillation, reactive power oscillation, voltage oscillation and frequency oscillation are extracted at the dominant frequency, and branch response points are generated by the degree of same direction of active power oscillation relative to frequency oscillation and the degree of same direction of reactive power oscillation relative to voltage oscillation.
[0010] The reflux closure is determined based on the degree of opposite direction and similar intensity of the response points of any two parallel branches, and the two parallel branches corresponding to the maximum non-zero reflux closure are determined as reflux branch pairs.
[0011] The basic active power reference and the basic reactive power reference are formed by conventional active power correction and conventional reactive power correction;
[0012] Active and reactive power pre-corrections are generated based on the active and reactive power oscillations of the return branch pairs.
[0013] The active power symmetry margin and reactive power symmetry margin are determined based on the active power limit value, reactive power limit value, basic active power reference value, and basic reactive power reference value at both ends of the return branch.
[0014] The active pre-correction and reactive pre-correction are limited to the same amplitude according to the active symmetry margin and reactive symmetry margin respectively, so as to obtain the active pair correction and reactive pair correction. The active pair correction and reactive pair correction are superimposed on the basic active reference and the basic reactive reference to obtain the final active reference and the final reactive reference, which are then sent to the strain gauge.
[0015] To achieve the above-mentioned objectives, the present invention provides the following technical solution.
[0016] Compared with the prior art, the present invention has the following substantial features and significant progress:
[0017] This invention, based on existing measurements of branch active power, branch reactive power, bus voltage, and bus frequency, determines the dominant power oscillation mode within a data window. It locates potentially canceling power swings at the common AC bus to the return branch pair and generates paired active and reactive power corrections within active and reactive power limits. This process allows the vibration suppression command to directly act on the two parallel branches generating the round-trip power, reducing branch power swings and branch current stress, while maintaining the algebraic sum of additional active and reactive power on the return branch pair at zero. Therefore, it does not alter the total active and reactive power corrections of the microgrid formed by conventional secondary control. Attached Figure Description
[0018] Figure 1 This is a diagram of the microgrid collaborative control architecture and mechanism based on power oscillation mode identification of the present invention. Detailed Implementation
[0019] The following description is provided in conjunction with the accompanying drawings. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the invention; those skilled in the art can make equivalent substitutions or combinations for the specific implementations.
[0020] This specific implementation describes a microgrid collaborative control method based on power oscillation mode identification, applied to a microgrid with multiple converters connected in parallel to a common AC bus. The microgrid includes multiple distributed power converters, energy storage converters, or load-side converters, each converter connected in parallel to the same common AC bus via its own AC branch.
[0021] like Figure 1 As shown, the parallel branch group 100 represents a controlled object in which multiple converters operate in parallel through their respective AC branches, and the common AC bus 200 represents the electrical node commonly connected by all the parallel branches. The measurement acquisition unit 300, the data window forming unit 400, the power oscillation mode identification unit 500, the return branch pair determination unit 600, the paired correction and limiting unit 700, and the converter reference transmission unit 800 sequentially constitute the control link from measurement to reference transmission.
[0022] Specifically, when two parallel branches exhibit opposite directions and similar intensities of power oscillation under the same dominant power oscillation mode, the combined oscillation at the common AC bus 200 may be partially canceled out. It is difficult to determine the two parallel branches primarily bearing the round-trip power by observing the combined quantity at the common AC bus 200 or the average quantity of adjacent units. Therefore, using branch active power, branch reactive power, bus voltage, and bus frequency as existing measurement inputs, the dominant power oscillation mode and dominant frequency are determined through power oscillation mode identification. Then, the branch response point and return current closure degree are formed at the dominant frequency, thereby identifying the return current branch pair. Equal-intensity and opposite-inverse active power pair corrections and reactive power pair corrections are generated for this return current branch pair.
[0023] The aforementioned control link identifies and suppresses the inter-branch round-trip power generated by each parallel branch under the same dominant power oscillation mode. First, the dominant frequency is determined to allow for comparison of the response direction and intensity of each parallel branch under the same oscillation component, avoiding the mixing of amplitudes and phases of different frequency components. The branch response point characterizes the response direction and intensity of each parallel branch at the dominant frequency. When two parallel branches have opposite response directions and similar response intensities, it indicates a strong correlation between the power increment of one parallel branch and the power decrement of the other at the dominant frequency; therefore, this pair of parallel branches is considered a candidate channel for inter-branch round-trip power.
[0024] Equal and opposite active and reactive paired corrections are applied to the defined return branch pairs, making the direction of the new corrections opposite to the half-difference of the return branch pairs at the dominant frequency, thus forming negative feedback on the opposite oscillations between the two parallel branches. Simultaneously, since the corrections at both ends of the return branch pairs are equal and opposite, the algebraic sum of the additional active and reactive power is zero, therefore it does not change the total active and reactive power correction targets of the microgrid already established by conventional secondary control.
[0025] Furthermore, the implementation method of the parallel branch group 100 and the common AC bus 200 is as follows:
[0026] The parallel branch group 100 consists of at least two parallel branches, each of which includes a strain gauge current transformer and its AC branch. Each parallel branch has independent active power, reactive power, upper active power limit, lower active power limit, upper reactive power limit, and lower reactive power limit. The common AC bus 200 provides a unified bus voltage and bus frequency reference for all parallel branches in the parallel branch group 100, enabling the active power and reactive power of each parallel branch to be compared at the same electrical node.
[0027] In one embodiment, the parallel branch group 100 includes three converters with the same or different capacities. Each converter maintains its branch output through a voltage loop, a current loop, and droop control, and receives conventional active power correction and conventional reactive power correction through secondary coordinated control. Two, three, or more parallel branches can be configured according to the actual scale of the microgrid, as long as each parallel branch can provide branch active power and branch reactive power, and can receive final active power reference and final reactive power reference, this specific embodiment can be implemented.
[0028] The common AC bus 200 provides a unified reference for bus voltage and bus frequency. For branch active power, the bus frequency reflects the active power balance of the common AC bus 200, and the degree to which the branch active power moves in the same direction relative to the frequency fluctuation reflects the direction of the active power frequency response of the parallel branch at the dominant frequency. For branch reactive power, the bus voltage reflects the reactive power and voltage balance of the common AC bus 200, and the degree to which the branch reactive power moves in the same direction relative to the voltage fluctuation reflects the direction of the reactive power voltage response of the parallel branch at the dominant frequency.
[0029] The relationship between the parallel branch group 100 and the common AC bus 200 enables this embodiment to identify the inter-branch round-trip power that is partially canceled at the common AC bus 200. When one parallel branch increases its active power at the dominant frequency, while another parallel branch decreases its active power by a similar magnitude at the same dominant frequency, the combined active power fluctuation at the common AC bus 200 may be small, but round-trip power still exists between the two parallel branches. Similarly, when the reactive power of two parallel branches fluctuates in opposite directions at the same dominant frequency, the combined reactive power or bus voltage fluctuation at the common AC bus 200 may not adequately reflect the branch current stress.
[0030] The upper and lower limits of active power represent the first... The active power limit, upper reactive power limit, and lower reactive power limit of the parallel branch allowable output of the branch represent the active power limit of the branch. The reactive power boundary of the parallel branch is allowed to be output. The above boundary can be derived from the converter nameplate capacity, energy storage status, temperature rise limit, protection setting value or the real-time allowable value of the controller.
[0031] Furthermore, the implementation methods of the measurement acquisition unit 300 and the data window forming unit 400 are as follows:
[0032] The measurement and acquisition unit 300 synchronously acquires the active and reactive power of each parallel branch, as well as the bus voltage and bus frequency of the common AC bus 200. Branch active power represents the instantaneous active power transferred from the parallel branches to the common AC bus 200, and branch reactive power represents the instantaneous reactive power transferred from the parallel branches to the common AC bus 200. Bus voltage represents the voltage amplitude at the common AC bus 200, and bus frequency represents the frequency at the common AC bus 200.
[0033] In the The nth synchronous sampling point, for the nth The parallel branch path obtains the branch path's contribution And the branch road was ineffective And obtain the bus voltage from the common AC bus 200. and bus frequency The active and reactive power of a branch can be calculated using the voltage and current sampling values of the corresponding parallel branches according to conventional power calculations. The bus voltage can be obtained by sampling the existing voltage of the voltage transformer or converter. The bus frequency can be obtained by the phase-locked loop or synchronous measurement unit at 200 on the common AC bus.
[0034] The data window forming unit 400 is used to form a data window. When forming the data window, the branch active power, branch reactive power, bus voltage, and bus frequency are made to use the same sampling interval and the same number of window points, and the branch active power, branch reactive power, bus voltage, and bus frequency are kept to have the same window start and end times. The sampling interval is denoted as... The number of window points is recorded as Within a data window, the sampling point numbers satisfy... Each type of measurement corresponds to the same set of sampling point numbers. Here, Ts represents the synchronous sampling interval used by the measurement acquisition unit 300, in seconds (s), and N represents the number of synchronous sampling points contained in a data window.
[0035] In the engineering implementation, the data window forming unit 400 pre-sets the proposed analysis frequency band, which is determined by the lower limit of the proposed analysis frequency band. and the upper limit of the frequency band to be analyzed Limited, and satisfy The sampling interval and the number of window points are constrained by the following formula to ensure that power oscillations within the frequency band to be analyzed can be resolved and that there is sufficient observation time:
[0036]
[0037]
[0038] The lower and upper limits of the proposed analysis frequency band can be determined based on microgrid small-signal oscillation tests, historical operating records, or existing protection waveform recording frequency bands. Therefore, each highest frequency cycle should have at least ten sampling points, and a data window should cover at least four cycles of the lowest frequency. For microgrids with common converters in parallel, the proposed analysis frequency band can be set in the frequency band where low-frequency power oscillations occur, and the sampling interval and the number of window points are jointly determined by the controller's computing power and control cycle.
[0039] The sampling interval Ts and the number of window points N are set according to the frequency band to be analyzed. When the upper limit of the frequency band to be analyzed is increased, Ts is reduced accordingly while satisfying the aforementioned constraint on the number of sampling points for the highest frequency period, to ensure that higher frequency oscillations have sufficient discrete sampling points. When the lower limit of the frequency band to be analyzed is decreased, N is increased accordingly or the data window observation time is extended while satisfying the aforementioned constraint on the number of coverage points for the lowest frequency period, to ensure that lower frequency oscillations have sufficient observation periods. Under the premise of satisfying the above constraints, the specific values of Ts and N are jointly determined by the controller's computing power and control cycle.
[0040] When forming a data window, the data window forming unit 400 performs missing point checks, outlier removal, and DC component removal on branch active power, branch reactive power, bus voltage, and bus frequency. It also performs bandpass filtering on the branch active power and branch reactive power for the target analysis frequency band. Missing point checks are used to detect sampling communication interruptions, short-term sampling channel failures, or timestamp discontinuities. If the absolute value of the timestamp interval between two adjacent sampling points deviates from the sampling interval by more than the allowable timestamp deviation, the corresponding sampling point is judged as a missing correlation point. The allowable timestamp deviation can be taken as a conventional synchronization tolerance not exceeding one-tenth of the sampling interval.
[0041] Outlier removal is used to eliminate sampling points that clearly exceed the physical boundaries or rate of change boundaries of the equipment. When the active power of a branch exceeds the extended boundaries defined by the upper and lower active power limits, the reactive power of a branch exceeds the extended boundaries defined by the upper and lower reactive power limits, the bus voltage exceeds the existing voltage protection boundary, the bus frequency exceeds the existing frequency protection boundary, or the difference between adjacent sampling points exceeds the allowable rate of change of the existing sampling channel, the corresponding sampling point is designated as an outlier. The proportion of outliers is denoted as [missing information]. The number of outliers is recorded as The proportion of outliers is obtained by the following formula:
[0042]
[0043] The upper limit of the proportion of outliers is taken as follows The upper limit for the number of consecutive outliers is denoted as And determine it according to the following formula:
[0044]
[0045] If outliers can be filled in by continuous sampling from the same sensor using either the method of maintaining the previous valid sample value or linear interpolation, then a data window will continue to be formed after filling in the outliers. If the proportion of outliers exceeds the upper limit of the outlier proportion, or the number of consecutive outliers exceeds the upper limit of the number of consecutive outliers and cannot be filled in by continuous sampling from the same sensor, then the current data window will not participate in the control update. The upper limit of the outlier proportion can be selected within the above range according to the measurement noise level. When the measurement noise level is high, a value closer to the lower limit should be selected to make the data window update more stringent.
[0046] DC component removal is achieved by subtracting the arithmetic mean of each measurement within the current data window from the measurement of each type within the data window. For any measurement sequence... Its oscillation sequence can be represented as:
[0047]
[0048] In the formula It represents any measurement sequence among branch active power, branch reactive power, bus voltage, or bus frequency. This represents the oscillation sequence after the DC component has been removed.
[0049] Bandpass filtering is used to preserve power oscillations within the target frequency band and reduce the impact of DC bias and high-frequency switching noise on the dominant power oscillation mode. Existing second- or fourth-order digital bandpass filters can be used. The lower cutoff frequency of the filter is taken as the lower limit of the target frequency band, and the upper cutoff frequency is taken as the upper limit of the target frequency band. All parallel branches use the same filter order, filter coefficients, and filter delay compensation. The filter coefficients can be obtained through bilinear transform, window function finite impulse response design, or existing digital filter tuning methods for the controller.
[0050] When any of the following measurements—branch active power, branch reactive power, bus voltage, and bus frequency—is missing within the data window and cannot be filled in by continuous sampling from the same sensor, the current data window will not participate in control updates. This avoids generating incorrect return branch pairs and incorrect active power pre-correction and reactive power pre-correction with incomplete data. For short-term single-point missing measurements, existing data can be filled in using methods such as maintaining the previous valid sample value or linear interpolation, and the filling should be limited to continuous sampling sequences from the same sensor.
[0051] The data window output by the data window forming unit 400 should include at least the data from each parallel branch. The active power sequence and reactive power sequence of the branches at each sampling point, and the common AC bus 200 at... The bus voltage sequence and bus frequency sequence are obtained at each sampling point. This output is simultaneously fed into the power oscillation mode identification unit 500 and the return branch pair determination unit 600. The data fed into the power oscillation mode identification unit 500 is used to determine the dominant power oscillation mode and dominant frequency, while the data fed into the return branch pair determination unit 600 is used to extract active power oscillation, reactive power oscillation, voltage oscillation, and frequency oscillation at the dominant frequency. The branch active power, after DC component removal processing and bandpass filtering of the proposed analysis band, is defined as the branch active power fitting sequence. The power oscillation mode identification unit 500 uses only the branch active power fitting sequence as the Prony fitting input; the branch active power fitting sequence has the same dimensions as the branch active power, but its steady-state components and components outside the frequency band to be analyzed have been filtered out.
[0052] Furthermore, the implementation of the power oscillation mode identification unit 500 is as follows:
[0053] The power oscillation mode identification unit 500 performs power oscillation mode identification on the active power of the branch to determine the dominant power oscillation mode and dominant frequency.
[0054] In one implementation, the power oscillation mode identification unit 500 performs Prony fitting on the active power of each parallel branch to obtain the swing amplitude, decay rate, swing frequency and initial phase of each oscillation component. The damping ratio is obtained based on the decay rate and swing frequency. The reconstruction result of each oscillation component is formed using the swing amplitude, decay rate, swing frequency and initial phase. The corresponding reconstruction energy is calculated based on the reconstruction result of each oscillation component. The oscillation component with the largest sum of reconstruction energy of the same candidate oscillation component in each parallel branch is determined as the dominant power oscillation mode.
[0055] Before performing Prony fitting, the power oscillation mode identification unit 500 uses the same fitting order for the active power fitting sequence of each parallel branch. The fitting order is denoted as... The number of components is denoted as Both satisfy The number of components starts from... The number of components is gradually increased until the reconstruction error of all parallel branches is no greater than the upper limit of reconstruction error, or until the maximum number of components preset by the controller is reached. The upper limit of reconstruction error can be set to... The maximum number of components can be taken as follows: For low-frequency power oscillations in parallel converter microgrids, the number of components is usually taken as a small integer that can cover the dominant low-frequency component and a small number of nearby disturbance components, so as to avoid fitting the switching noise with too high an order.
[0056] The upper limit of reconstruction error and the maximum number of components are used to constrain the balance between fitting accuracy and computational cost. Decreasing the upper limit of reconstruction error requires increasing the number of oscillating components to achieve higher reconstruction accuracy, but this correspondingly increases computational cost and the likelihood of fitting noise components. Conversely, increasing the upper limit of reconstruction error reduces the required number of oscillating components, but decreases the fitting accuracy for minor components other than the dominant oscillating components. The maximum number of components limits the Prony fitting order; its specific value is determined by ensuring coverage of the dominant low-frequency components and a small number of nearby disturbance components within the analyzed frequency band, while allowing the controller to complete the calculation within one control cycle.
[0057] For the first Establish a linear prediction relationship for the active power oscillation sequence of parallel branches:
[0058]
[0059] This formula is in It holds true at the sampling points, and the unknown quantity The linear prediction coefficients are obtained using the least squares method. After obtaining the linear prediction coefficients, the power oscillation characteristic polynomial is formed and its roots are found. The power oscillation characteristic polynomial can be expressed as:
[0060]
[0061] in Let represent the characteristic root variable, whose values are obtained by finding the roots of the power oscillation characteristic polynomial. For the ... The first parallel branch A complex root The decay rate and oscillation frequency are obtained by the following formula:
[0062]
[0063]
[0064] in The roots of the power oscillation characteristic polynomial are... Indicates the rate of decay. This represents the oscillation frequency. For a measured power sequence, conjugate roots correspond to the same oscillation component. The power oscillation mode identification unit 500 forms the oscillation amplitude and initial phase according to the positive frequency component. The oscillation amplitude and initial phase can be obtained by substituting the sine and cosine basis functions into the least squares equation.
[0065] For the The first parallel branch The active power of the fitted oscillation component is:
[0066]
[0067] in Indicates the first Parallel branch road The oscillation component at the th oscillation component in ... The fitting of each sampling point is effective. Indicates the amplitude of the swing. Indicates the rate of decay. Indicates the oscillation frequency. The initial phase is indicated. The sampling interval is given by the synchronous sampling clock of the measurement acquisition unit 300, and the swing amplitude, decay rate, swing frequency, and initial phase are all obtained by the Prony fitting process described above.
[0068] The decay rate obtained from the Prony fitting is used to characterize how quickly the corresponding oscillation component decays or grows over time; the oscillation frequency is used to characterize the change in the oscillation period per unit time; the oscillation amplitude is used to characterize the strength of the oscillation component in the corresponding parallel branch; and the initial phase is used to determine the phase position of the oscillation component at the beginning of the data window. The fitting function formed by the above parameters is used to reconstruct the time-domain change process of the corresponding oscillation component within the current data window.
[0069] For the first When the active power of a parallel branch is reconstructed using multiple components, the reconstructed result of the fitted active power is:
[0070]
[0071] The reconstruction error can be calculated using the following formula:
[0072]
[0073] If the sum of squares of the active power fitted sequence of the branch corresponding to the denominator is less than the lower limit of the sum of squares of active power oscillation, then the parallel branch is not considered as a candidate contributing branch of the dominant power oscillation mode in the current data window. By constraining the reconstruction error and the lower limit of the sum of squares of active power oscillation, the power oscillation mode identification unit 500 can avoid misjudging the active power of branches with pure noise or almost no oscillation as the dominant power oscillation mode.
[0074] The damping ratio is used to verify the decay state of the oscillating component and is obtained by the following formula:
[0075]
[0076] The damping ratio is dimensionless, the decay rate is measured in seconds, and the oscillation frequency is measured in Hertz. The damping ratio varies with the decay rate and oscillation frequency, and is used to determine whether the dominant power oscillation mode has improved in the next data window.
[0077] The oscillation frequencies obtained by Prony fitting on different parallel branches may have slight deviations. The power oscillation mode identification unit 500, based on the principle of frequency proximity, groups oscillation components within the same frequency tolerance into the same candidate oscillation component, and calculates the reconstruction energy for the reconstruction results of the same candidate oscillation component across all parallel branches. The frequency tolerance is denoted as... It can be determined by the following formula:
[0078]
[0079] in Indicates the frequency resolution of the data window. This represents a 2% frequency margin of the dominant frequency; taking the larger of these two values helps avoid incorrect splitting of the same dominant power oscillation mode due to discretization resolution. The reconstruction energy of each candidate oscillation component is:
[0080]
[0081] The reconstructed energy is used to sort different candidate oscillation components.
[0082] The reconstructed energy is obtained by accumulating the reconstruction results of the corresponding candidate oscillation component within the current data window, and is used to characterize the strength of the contribution of the candidate oscillation component to the active power oscillation of the branch. By summarizing the reconstruction energy of the same candidate oscillation component in all parallel branches, it is possible to avoid selecting the dominant mode based solely on the local oscillation of a single parallel branch. The larger the summed reconstruction energy, the greater the overall contribution of the candidate oscillation component to the power oscillation of the current parallel system branch, and therefore it is regarded as the dominant power oscillation mode.
[0083] The dominant number can be represented as:
[0084]
[0085] With dominant number The corresponding oscillation component is the dominant power oscillation mode. The dominant frequency can be taken as the average of the oscillation frequencies of each parallel branch belonging to the dominant power oscillation mode, weighted by the reconstructed energy:
[0086]
[0087] in Indicates the dominant frequency. Indicates the first The branch reconstruction energy of the parallel branches for the dominant power oscillation mode. If the denominator is less than the lower limit of the dominant frequency weighted energy, the current data window will not output the dominant power oscillation mode, and the generation of subsequent return branch pairs for determination and pairwise correction will be stopped. The lower limit of the dominant frequency weighted energy has the same dimensions as the branch reconstruction energy, and can be taken as the sum of the lower limits of the squares of the active power swing of all parallel branches multiplied by the sampling interval.
[0088] Furthermore, the implementation of the return branch for determining unit 600 is as follows:
[0089] The return branch pair determination unit 600 receives the dominant power oscillation mode and dominant frequency output by the power oscillation mode identification unit 500, and receives the same data window output by the data window forming unit 400. After the dominant frequency is given, the return branch pair extracts the active power oscillation, reactive power oscillation, voltage oscillation, and frequency oscillation from the branch active power, branch reactive power, bus voltage, and bus frequency in the same data window.
[0090] To extract the four types of oscillations at the dominant frequency, the return branch pairs the determination unit 600 with any measurement sequence to be extracted. Construct sine and cosine basis functions corresponding to the dominant frequency, and use least squares projection to obtain the oscillation component at the dominant frequency:
[0091]
[0092] in Represents measurement sequence The oscillation component at the dominant frequency and These are the least squares projection coefficients. For active power in the branch, take... Obtaining active oscillation For branch roads that are ineffective, take Obtaining reactive oscillation For bus voltage, take Voltage swing For the bus frequency, take Receive frequency oscillation The four types of oscillations mentioned above are all extracted from data windows with the same sampling interval, the same number of window points, and the same start and end times of the window. Therefore, the phase comparison has the same time reference.
[0093] After extracting the four types of oscillations at the dominant frequency, the return branch calculates the degree of symmetry between the active power oscillation and the frequency oscillation, and the degree of symmetry between the reactive power oscillation and the voltage oscillation, respectively, for the determining unit 600. The degree of active power in the same direction for a parallel branch can be expressed as:
[0094]
[0095] The degree of reactive power homing can be expressed as:
[0096]
[0097] To avoid division by zero or near-zero, the return branch sets lower limits for the sum of squares of active power swing, reactive power swing, voltage swing, and frequency swing for the determining unit 600. For the first... The lower limit of the sum of squares of the active power oscillation in a parallel branch can be expressed as:
[0098]
[0099] The lower limit of the sum of squares of reactive oscillation can be expressed as:
[0100]
[0101] The lower limit of the sum of squares of voltage swings can be expressed as:
[0102]
[0103] The lower limit of the sum of squares of frequency oscillations can be expressed as:
[0104]
[0105] in Indicates the first The rated active power of the parallel branches, Indicates the first The rated reactive power of the parallel branches, Indicates the rated bus voltage. Indicates the rated bus frequency; , , and These represent the active power swing determination coefficient, reactive power swing determination coefficient, voltage swing determination coefficient, and frequency swing determination coefficient, respectively, and can all be taken as... When the sum of squares of active power swing is lower than the lower limit of the sum of squares of active power swing, or the sum of squares of frequency swing is lower than the lower limit of the sum of squares of frequency swing, the degree of active power unidirectionality is recorded as zero; when the sum of squares of reactive power swing is lower than the lower limit of the sum of squares of squares of reactive power swing, or the sum of squares of voltage swing is lower than the lower limit of the sum of squares of voltage swing, the degree of reactive power unidirectionality is recorded as zero.
[0106] The active and reactive oscillation degrees are used to compare the response direction at the same dominant frequency. Since all four types of oscillations are extracted from the same data window and the same dominant frequency, their phases share a common time reference. When calculating the oscillation degree using the normalized inner product form, the sign of the oscillation degree is used to characterize the same or opposite relationship between the two types of oscillations, and its absolute value is used to characterize the degree of significance of the directional relationship. Normalization is used to reduce the influence of different measurement amplitude scales on the direction judgment.
[0107] The aforementioned active power swing determination coefficient, reactive power swing determination coefficient, voltage swing determination coefficient, and frequency swing determination coefficient are used to form the corresponding lower limit of the swing square sum, so as to eliminate the influence of normal measurement noise, switching ripple, and steady-state micro-fluctuations on the calculation of the degree of unidirectionality. In specific settings, when there are no obvious low-frequency power oscillations that need to be suppressed in the microgrid, background operating data can be collected according to the same sampling interval, number of window points, and filtering method as the actual control. Based on the background fluctuation levels of branch active power, branch reactive power, bus voltage, and bus frequency in the frequency band to be analyzed, the corresponding determination coefficients are determined within the aforementioned value range, so that the lower limit of the swing square sum formed by the determination coefficients is higher than the normal background fluctuation level and lower than the minimum effective oscillation level that needs to be identified.
[0108] When the background noise of a measurement channel increases and causes low-energy fluctuations to be misjudged as valid oscillations, the corresponding oscillation determination coefficient increases within the aforementioned range. Conversely, when the measurement noise is low and it is necessary to improve the sensitivity to weak low-frequency oscillations, the corresponding oscillation determination coefficient decreases within the aforementioned range. Active and reactive power oscillations are measured using the rated active and rated reactive power of the corresponding branches, respectively, while voltage and frequency oscillations are measured using the rated bus voltage and rated bus frequency, respectively. This allows for tuning of different capacity branches and different measurement channels using dimensionless determination coefficients. These determination coefficients remain constant within the same data window and are readjusted when the measurement noise level, sampling conditions, or filtering conditions change continuously.
[0109] Branch response points are comparable objects with direction and intensity. In one implementation, the first... The branch response points of a parallel branch can be represented as:
[0110]
[0111] in Indicates the branch response point, This represents the reactive power weighting coefficient. The reactive power weighting coefficient is a dimensionless quantity and satisfies... Default value When the voltage control priority of the common AC bus 200 is higher than the frequency control priority, the reactive power weighting factor can be taken according to the voltage control priority factor in the existing controller. When frequency control priority is higher than voltage control priority, the reactive power weighting factor can be taken from the frequency control priority factor in the existing controller. The reactive power weighting coefficient is updated only within the controller tuning cycle and remains unchanged within the same data window.
[0112] The direction of a branch response point is indicated by its sign, and the strength of the branch response point is indicated by its absolute value. For any two parallel branches... and The degree of opposite direction can be expressed as:
[0113]
[0114] The degree of similarity in strength can be expressed as:
[0115]
[0116] in This represents the lower limit of the branch response point intensity, used to avoid the denominator being zero. The lower limit of the branch response point intensity is a dimensionless quantity and can be taken as... Default value .like or If the strength is below the lower limit of the branch response point, the return closure of the corresponding two parallel branches is recorded as zero. Strength close to the tolerance is recorded as... It is acceptable Default value If the difference in strength similarity between two sets of parallel branches is less than the strength similarity tolerance, then the parallel branch pair with greater opposite direction is preferred; if the opposite direction is still the same, then the parallel branch pair with the larger sum of the reconstruction energies of the dominant frequency branches of the two parallel branches is selected; if they are still the same, then they are selected according to the order of the parallel branch numbers. This rule provides the boundary for strength similarity and the method for determining the parallel case.
[0117] The lower limit of branch response point strength is used to exclude branch response points whose absolute value is close to zero and are mainly affected by measurement noise or numerical error. It can be determined within the aforementioned range based on the background fluctuation level of the branch response point under operating conditions without significant low-frequency power oscillations. A larger value is taken when the background fluctuation increases, and a smaller value is taken when it is necessary to improve the sensitivity of weak oscillation identification and the background fluctuation is small. The strength proximity tolerance is used to determine whether the strength proximity of different parallel branch pairs can be considered equivalent. It can be determined within the aforementioned range based on the fluctuation range of strength proximity caused by measurement noise, window discrepancies, and fitting errors in historical operation or small disturbance tests, and remains unchanged within the same data window.
[0118] The return closure increases with the degree of opposition between the branch response points of the two parallel branches, and also with the degree of proximity in strength between the branch response points of the two parallel branches. The return closure can be expressed as:
[0119]
[0120] in Indicates the first Parallel branch and the first The return closure degree between parallel branches. The degree of opposite direction, similarity of intensity, and branch response points are all dimensionless; therefore, the return closure degree is dimensionless. If any branch response point has no comparable intensity, the corresponding return closure degree is recorded as zero. If the maximum return closure degree is non-zero, the two parallel branches corresponding to this maximum non-zero return closure degree are determined as a return branch pair; if the return closure degree corresponding to any two parallel branches is zero, no return branch pair is set in the current data window, and the steps for generating active power pre-correction and reactive power pre-correction are stopped. The current control cycle only retains the basic active power reference and basic reactive power reference.
[0121] The degree of opposite direction reflects whether the response directions of two parallel branches are opposite at the dominant frequency, while the degree of similarity in intensity reflects whether the response intensities of two parallel branches are similar. Return closure combines these two conditions for branch pair comparison: only when the response directions of two parallel branches are relatively opposite and their response intensities are relatively similar does the corresponding return closure achieve a large value. Therefore, the parallel branch pair corresponding to the maximum non-zero return closure can be used to characterize the main inter-branch round-trip power path within the current data window.
[0122] Furthermore, the implementation of the paired correction limiting unit 700 is as follows:
[0123] In the presence of a return current branch pair, the paired correction limiting unit 700 takes the active power half-difference of the two active power swings of the return current branch pair and the reactive power half-difference of the two reactive power swings of the return current branch pair. Assume the return current branch pair is composed of the first... Parallel branch and the first Composed of several parallel branches, the active half-difference and reactive half-difference can be expressed as follows:
[0124] Taking half-differences for the two active and two reactive oscillations of the return current branch pair is to characterize the opposite oscillation components of the two parallel branches relative to their common intermediate state. When the two parallel branches exhibit approximately equal-amplitude, opposite-opposite oscillations at the dominant frequency, the half-difference can retain the main components with opposite changes at both ends. Multiplying this half-difference by an opposite-direction damping gain forms a pre-correction, and applying corrections of opposite signs to both ends of the two parallel branches creates directional negative feedback for the opposite oscillations of the return current branch pair.
[0125]
[0126]
[0127] The paired correction limiting unit 700 uses active power vibration suppression gain to convert the active power half-difference into active power pre-correction, and uses reactive power vibration suppression gain to convert the reactive power half-difference into reactive power pre-correction.
[0128]
[0129]
[0130] Active vibration damping gain and reactive power vibration damping gain All are dimensionless quantities and satisfy the following conditions: , The above-mentioned value range ensures that the pre-correction direction is opposite to the round-trip power oscillation at the dominant frequency and avoids applying an additional reference exceeding half the difference amplitude within a single data window. During actual tuning, the active and reactive power suppression gains can be determined by existing controllers through small-disturbance tests, historical oscillation attenuation records, or stability margin checks, and remain constant within a single data window.
[0131] Simultaneously with the generation of active power pre-correction and reactive power pre-correction, the existing secondary control generates conventional active power correction and conventional reactive power correction based on frequency recovery, voltage recovery, and active and reactive power sharing. The basic active power reference is obtained by superimposing the original active power reference with the conventional active power correction, and the basic reactive power reference is obtained by superimposing the original reactive power reference with the conventional reactive power correction. In this specific embodiment, the paired active power correction and paired reactive power correction are only superimposed on both ends of the return branch pair.
[0132] The paired correction limiting unit 700 determines the active power symmetry margin based on the upper and lower active power limits and the basic active power reference at both ends of the return branch, and determines the reactive power symmetry margin based on the upper and lower reactive power limits and the basic reactive power reference at both ends of the return branch. For the first... Parallel branch and the first The active power margin of a parallel branch can be expressed as:
[0133]
[0134] The reactive power symmetry margin can be expressed as:
[0135]
[0136] in and This indicates the active power reference of the return branch to the foundations at both ends. and This indicates the reactive power reference of the return branch to the foundations at both ends. and These represent the upper limit and lower limit of active power, respectively. and These represent the upper and lower limits of reactive power, respectively. Through the calculation of the outer maximum value, when any basic active or reactive power reference is outside the boundary, resulting in a negative symmetry margin, the corresponding symmetry margin is set to zero, thus avoiding the formation of a same-amplitude correction that conflicts with the boundary.
[0137] The active power symmetry margin and reactive power symmetry margin are respectively taken from the common allowable value of the available adjustment margin at both ends of the return branch, and the common allowable value is used to perform the same amplitude limiting on the pre-correction at both ends. Thus, when either end approaches the active power boundary or the reactive power boundary, the correction amount at both ends is limited synchronously, avoiding the situation where one end is cut off while the other end still maintains the original correction amount due to independent limiting at both ends. This ensures that the correction amount of the return branch at both ends is equal and opposite, and at the same time meets the allowable range of the equipment at both ends.
[0138] The paired correction limiting unit 700 performs same-amplitude limiting on the active power pre-correction according to the active power symmetry margin, and performs same-amplitude limiting on the reactive power pre-correction according to the reactive power symmetry margin, thus obtaining paired active power correction and paired reactive power correction:
[0139]
[0140]
[0141] If the active power symmetry margin is zero, then the active power paired correction is zero; if the reactive power symmetry margin is zero, then the reactive power paired correction is zero. After limiting, the first... Parallel branch superposition and , No. Parallel branch superposition and Alternatively, the positive and negative directions can be interchanged based on the signs of the active and reactive half-differences, ensuring that the correction direction is always opposite to the swing half-difference of the return branch at the dominant frequency. Since the corrections at both ends are equal in magnitude and opposite in direction, the algebraic sum of the additional active power after limiting is zero, and the algebraic sum of the additional reactive power after limiting is also zero.
[0142] Furthermore, the implementation method for the converter reference distribution unit 800 and the next data window review is as follows:
[0143] The converter reference sending unit 800 superimposes the active power pair correction and reactive power pair correction onto the base active power reference and base reactive power reference at both ends of the return branch pair to obtain the final active power reference and final reactive power reference at both ends of the return branch pair. For parallel branches not identified as return branch pairs, the final active power reference and final reactive power reference remain the corresponding base active power reference and base reactive power reference. After the final active power reference and final reactive power reference are sent to the corresponding converter, the primary controller adjusts the AC side voltage phase and amplitude through the existing voltage loop, current loop, droop primary control, or power point tracking control to make the actual power of the branch follow the reference value.
[0144] The converter reference sending unit 800 can send the final active power reference and final reactive power reference to the corresponding converter via analog interface, digital communication bus, or internal controller variable writing. The sent data should include the parallel branch number, final active power reference, final reactive power reference, control cycle timestamp, data validity flag, and verification field; the verification field uses the cyclic redundancy check or frame check generation rules in the existing communication protocol. The refresh period for the final active power reference and final reactive power reference is equal to the control cycle. Each time the data window completes an update, a reference distribution is triggered. The distribution timeout can be expressed as:
[0145]
[0146] If the data sent from the other end of any return branch is not acknowledged by the corresponding current transformer within the timeout period, or if the verification field fails, the data with the same control cycle timestamp will be resent in the next control cycle. If the resentment still fails, the corresponding parallel branch will not update the active power pair correction and reactive power pair correction, and will maintain the basic active power reference and basic reactive power reference according to the existing controller safety policy, or perform a smooth transition on the previous valid final active power reference and the previous valid final reactive power reference.
[0147] After issuing the final active and reactive power references, the dominant power oscillation mode, dominant frequency, and return branch pair are redefined in the next data window. To avoid misjudgments due to random noise, the dominant frequency of the current data window is compared with the dominant frequency of the previous data window during the review in the next data window. The dominant frequency tracking tolerance is denoted as... It can be determined by the following formula:
[0148]
[0149] If the difference between the two values does not exceed the dominant frequency tracking tolerance, they are considered to correspond to the same dominant power oscillation mode; if the difference exceeds the dominant frequency tracking tolerance, the current dominant power oscillation mode is taken as the new oscillation object, and the verification is restarted with the current data window.
[0150] When the same dominant power oscillation mode satisfies the conditions of reduced reconstruction energy and improved damping ratio in the next data window, the dominant power oscillation mode is considered to have improved. This criterion can be expressed as:
[0151]
[0152]
[0153] in Indicates the first The dominant power oscillation mode reconfiguration energy of each data window. Indicates the first The dominant power oscillation mode damping ratio for each data window. This indicates that reconfiguration energy improves tolerance. This indicates that the damping ratio improves the tolerance. The reconstruction energy improves the tolerance, which is a dimensionless quantity and can be taken as... Default value The damping ratio improvement tolerance is a dimensionless quantity, which can be taken as... Default value The preset number of confirmation windows is denoted as... It is acceptable Default value The reconfiguration energy improvement tolerance, damping ratio improvement tolerance, and preset confirmation window number are used to avoid misjudging small fluctuations in values as improvements. If the reconfiguration energy does not decrease or the damping ratio does not improve, the active power pair correction and reactive power pair correction will continue to be updated; if the reconfiguration energy decreases and the damping ratio improves, and the preset confirmation window number is reached continuously, the active power pair correction and reactive power pair correction will be smoothly returned to zero.
[0154] Reconstructed energy reflects the oscillation intensity of the dominant power oscillation mode within the data window, while the damping ratio reflects the decay state of this oscillation mode. Using both the reduction in reconstructed energy and the improvement in the damping ratio as verification conditions simultaneously can prevent the oscillation from being mistakenly judged as having stabilized and improved simply because the swing amplitude has decreased briefly within a certain data window. Smoothing back to zero is performed only after a preset number of confirmation windows have been reached to further reduce the impact of random measurement fluctuations on the exit from the vibration suppression correction judgment.
[0155] Smooth return to zero can be achieved by decreasing the control cycle one by one. Let the first... The active pair correction for each control cycle is The reactive pair correction is The zero-return coefficient is ,and The smooth return-to-zero process can then be expressed as:
[0156]
[0157]
[0158] When the absolute value of the active power pairwise correction is lower than the minimum active power adjustment, the active power pairwise correction is set to zero; when the absolute value of the reactive power pairwise correction is lower than the minimum reactive power adjustment, the reactive power pairwise correction is set to zero. The zero-return coefficient, the minimum active power adjustment, and the minimum reactive power adjustment can be set using the conventional settings in the existing converter reference smoothing process.
[0159] As an optional implementation, multiple converters in the parallel branch group 100 jointly supply power to the common AC bus 200. The branch active and reactive power vary slightly around the baseline active and reactive power references, while the bus voltage and bus frequency meet existing operating boundaries. The measurement and acquisition unit 300 continues to operate according to the sampling interval. The system synchronously collects branch active power, branch reactive power, bus voltage, and bus frequency, and the data window forming unit 400 still forms a data window. The power oscillation mode identification unit 500 can identify low-energy oscillation components, but if the reconstruction energy of the dominant power oscillation mode is lower than the lower limit of the dominant frequency weighted energy, or if the return current closure calculated by the return current branch pair determination unit 600 is zero, then no return current branch pair is set in the current data window. At this time, the paired correction limiting unit 700 does not generate active power pre-correction and reactive power pre-correction, and the final active power reference and final reactive power reference issued by the converter reference issuing unit 800 are equal to the basic active power reference and basic reactive power reference.
[0160] As an alternative implementation, assume the first Parallel branch and the first The parallel branches exhibit active power oscillations with opposite directions and similar in intensity at the same dominant frequency, while their reactive power oscillations also show opposite directions relative to the voltage oscillations. The combined oscillation at the common AC bus 200 is partially canceled out by the opposite oscillations of the two parallel branches, making it difficult to determine the main round-trip power path based solely on the combined amount at the common AC bus 200. The power oscillation mode identification unit 500 determines the dominant power oscillation mode and dominant frequency based on the branch active power. The return branch pair determination unit 600 extracts the active power oscillation, reactive power oscillation, voltage oscillation, and frequency oscillation from the same data window at the dominant frequency, calculates the branch response point and return current closure, and identifies the two parallel branches corresponding to the maximum non-zero return current closure as the return branch pair. The paired correction limiting unit 700 then generates active power pre-correction and reactive power pre-correction based on the active power half-difference and reactive power half-difference, and obtains active power paired correction and reactive power paired correction through active power symmetry margin and reactive power symmetry margin. When the final active and final reactive references are applied to the strain gauge, the two parallel branches receive directional negative feedback opposite to the oscillation of their dominant frequency, and the round-trip power between the branches is weakened.
[0161] As an alternative implementation, it is assumed that the return branch pairs are already determined, but the basic active power reference of one parallel branch is close to the active power upper limit, or the basic reactive power reference of the other parallel branch is close to the reactive power lower limit. Directly applying active power pre-correction or reactive power pre-correction may cause the parallel branches to exceed the active power upper limit, active power lower limit, reactive power upper limit, or reactive power lower limit. The paired correction limiting unit 700 limits the paired active power correction and paired reactive power correction through active power symmetry margin and reactive power symmetry margin, keeping the correction amounts at both ends equal and opposite, and ensuring that the final active power reference and the final reactive power reference fall within the allowable range of the equipment. When the active power symmetry margin is zero, the paired active power correction is zero; when the reactive power symmetry margin is zero, the paired reactive power correction is zero. Under this condition, even if the directional vibration suppression amount is limited by the equipment boundary, the algebraic sum of the additional active power and the algebraic sum of the additional reactive power are still zero.
[0162] As an optional implementation, if the active power sampling of a parallel branch is interrupted, or if the bus frequency sampling of the common AC bus 200 is missing within the data window and cannot be supplemented by continuous sampling from the same sensor, the data window forming unit 400 determines that the current data window will not participate in the control update. In this case, the power oscillation mode identification unit 500 does not update the dominant power oscillation mode based on the data window, the return branch pair determination unit 600 does not update the return branch pair based on the data window, and the pairwise correction limiting unit 700 does not generate new active power pre-correction and reactive power pre-correction. The converter reference issuing unit 800 retains the basic active power reference and basic reactive power reference formed by conventional secondary control, or maintains a smooth transition of the previous effective final active power reference and the previous effective final reactive power reference according to the existing controller safety strategy.
[0163] As an optional implementation, the final active power reference and final reactive power reference have been issued to the corresponding converter. After the next data window is formed, the power oscillation mode identification unit 500 re-determines the dominant power oscillation mode and dominant frequency, and the return branch pair determination unit 600 re-determines the return branch pair. If the dominant frequency is still within the dominant frequency tracking tolerance, and the reconstruction energy of the dominant power oscillation mode has not decreased or the damping ratio has not improved, it indicates that the current directional negative feedback still needs to continue to operate. The pairwise correction limiting unit 700 continues to update the active power pairwise correction and reactive power pairwise correction according to the latest active power swing and reactive power swing. If the dominant frequency is still within the dominant frequency tracking tolerance, and the reconstruction energy of the dominant power oscillation mode has decreased and the damping ratio has improved, and the preset confirmation window number has been continuously reached, it indicates that the dominant power oscillation of the return branch pair has been suppressed. The converter reference issuing unit 800 smoothly returns the active power pairwise correction and reactive power pairwise correction to zero, so that the microgrid returns to the operating state dominated by conventional secondary control.
[0164] It should be noted that this specific implementation can be achieved by a centralized microgrid controller, a collection of distributed secondary controllers, or an equivalent control system formed by the local controllers of each converter and a communication network. The hardware carrier of the controller can be an industrial controller, an embedded controller, a digital signal processor, or a microgrid energy management and control platform with real-time sampling and communication capabilities.
[0165] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and all such modifications and variations fall within the scope defined by the appended claims.
Claims
1. A microgrid cooperative control method based on power oscillation mode identification, applied to a microgrid with multiple converters connected to a common AC bus, characterized in that, include: The active power and reactive power of each parallel branch, as well as the bus voltage and bus frequency, are collected synchronously to form a data window; Power oscillation mode identification is performed on the active power of the branch to determine the dominant power oscillation mode and dominant frequency; Active power oscillation, reactive power oscillation, voltage oscillation and frequency oscillation are extracted at the dominant frequency, and branch response points are generated by the degree of same direction of active power oscillation relative to frequency oscillation and the degree of same direction of reactive power oscillation relative to voltage oscillation. The reflux closure is determined based on the degree of opposite direction and similar intensity of the response points of any two parallel branches, and the two parallel branches corresponding to the maximum non-zero reflux closure are determined as reflux branch pairs. The basic active power reference and the basic reactive power reference are formed by conventional active power correction and conventional reactive power correction; Active and reactive power pre-corrections are generated based on the active and reactive power oscillations of the return branch pairs. The active power symmetry margin and reactive power symmetry margin are determined based on the active power limit value, reactive power limit value, basic active power reference value, and basic reactive power reference value at both ends of the return branch. According to the active symmetry margin and reactive symmetry margin, the active pre-correction and reactive pre-correction are limited to the same amplitude to obtain active pair correction and reactive pair correction. The active pair correction and reactive pair correction are then superimposed on the basic active reference and basic reactive reference to obtain the final active reference and final reactive reference, which are then sent to the strain gauge. Specifically, when generating the branch response point, the degree of symmetry between the active power swing and the frequency swing and the degree of symmetry between the reactive power swing and the voltage swing are combined into a comparable object with direction and intensity, and this comparable object with direction and intensity is used as the branch response point. Specifically, when determining the return closure degree, the return closure degree is increased as the direction of the response points of the two parallel branches increases, and as the intensity of the response points of the two parallel branches increases; when any of the response points of the branches does not have comparable intensity, the return closure degree related to that response point is recorded as zero.
2. The microgrid collaborative control method based on power oscillation mode identification according to claim 1, characterized in that, When forming the data window, the active power of the branch, the reactive power of the branch, the bus voltage, and the bus frequency are made to use the same sampling interval and the same number of window points, and the active power of the branch, the reactive power of the branch, the bus voltage, and the bus frequency are kept to have the same window start and end times.
3. The microgrid collaborative control method based on power oscillation mode identification according to claim 1, characterized in that, When forming the data window, the active power of the branch, the reactive power of the branch, the bus voltage, and the bus frequency are checked for missing points, eliminated for outliers, and processed to remove DC components. The active power and reactive power of the branch are also bandpass filtered for the frequency band to be analyzed. When any measurement of the active power, reactive power, bus voltage, and bus frequency is missing in the data window and cannot be filled in by continuous sampling from the same sensor, the current data window does not participate in the control update.
4. The microgrid collaborative control method based on power oscillation mode identification according to claim 1, characterized in that, When determining the dominant power oscillation mode and the dominant frequency, Prony fitting is performed on the active power of each parallel branch to obtain the swing amplitude, decay rate, swing frequency and initial phase of each oscillation component. The damping ratio is obtained according to the decay rate and the swing frequency. The reconstruction result of each oscillation component is formed using the swing amplitude, the decay rate, the swing frequency and the initial phase. The oscillation component with the largest sum of the reconstructed energies is determined as the dominant power oscillation mode.
5. The microgrid collaborative control method based on power oscillation mode identification according to claim 1, characterized in that, When the return closure degree corresponding to any two parallel branches is zero, the current data window does not set the return branch pair and stops executing the steps of generating the active power pre-correction and the reactive power pre-correction.
6. The microgrid collaborative control method based on power oscillation mode identification according to claim 1, characterized in that, When generating the active pre-correction and the reactive pre-correction for the return branch pair, the active half-difference is taken for the two active swings of the return branch pair, and the reactive half-difference is taken for the two reactive swings of the return branch pair. The active half-difference and the reactive half-difference are then converted into the active pre-correction and the reactive pre-correction using the active vibration suppression gain and the reactive vibration suppression gain, respectively.
7. A microgrid collaborative control method based on power oscillation mode identification according to claim 6, characterized in that, When performing the same amplitude limiting, the active pre-correction is limited according to the active symmetry margin to obtain the active paired correction, and the reactive pre-correction is limited according to the reactive symmetry margin to obtain the reactive paired correction, so that the algebraic sum of the additional active quantity after limiting is zero and the algebraic sum of the additional reactive quantity after limiting is zero.
8. The microgrid collaborative control method based on power oscillation mode identification according to claim 4, characterized in that, After issuing the final active power reference and the final reactive power reference, the dominant power oscillation mode, the dominant frequency, and the return branch pair are redefined in the next data window. If the reconstruction energy of the dominant power oscillation mode does not decrease or the damping ratio does not improve, the active power pair correction and the reactive power pair correction are updated. If the reconstruction energy of the dominant power oscillation mode decreases and the damping ratio improves, the active power pair correction and the reactive power pair correction are smoothly returned to zero.
Citation Information
Patent Citations
Multi-source time domain cooperative control method, system and device for virtual synchronous generator
CN122292556A
Self-adaptive rectification control method for grid-building oscillator of hybrid MMC low-frequency weak power grid
CN122393965A
