A method and device for obtaining time-varying inertia of wind farms based on cluster aggregation method

Abstract

The present invention discloses a method and device for obtaining the time-varying inertia of a wind farm station based on a clustering and aggregation method, belonging to the field of wind farm control technology. The method derives the frequency and reactive power of any wind turbine in the current sampling period as the equivalent frequency and equivalent power of the unit to which the wind turbine belongs; then, the equivalent frequency and equivalent power of each unit in the current sampling period and the two previous sampling periods are converted into two linear operators, and the equivalent inertia corresponding to each unit in the current sampling period is calculated using the linear operators corresponding to each unit; finally, the inertia level of the entire wind farm station is determined through weighted calculation. That is, the present invention takes into account the differences among internal units, clusters and aggregates units with similar inertial responses, estimates the inertia of each aggregated unit separately, and then determines the inertia level of the entire wind farm station through weighted calculation; this method has low computational complexity and can accurately reflect the influence of parameter distribution and estimate the time-varying inertia of the wind farm.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind farm control, and more specifically, relates to a method and device for obtaining time-varying inertia of a wind farm station based on a clustering aggregation method. Background Art

[0002] The increasing penetration of wind power in power grids has also raised numerous new safety and stability issues. Among them, the decline in inertia is one of the main factors limiting its large-scale application. Wind turbines are primarily integrated into the grid through power electronic converters, resulting in frequency decoupling from the AC power system and an inability to actively respond to system frequency changes. Therefore, the large-scale integration of wind power may lead to a decrease in the power system's inertia, weakening the system's interference mitigation and frequency stability. Furthermore, with the construction and operation of large-scale AC and DC power grids, the risk of large-scale disturbances increases. The support provided by interconnected DC asynchronous grids is relatively low, complicating frequency regulation control and inertia assessment. Furthermore, due to natural constraints, the geographical distribution of wind farms exacerbates the uneven temporal and spatial distribution of power system inertia. Therefore, to fully understand the temporal trends and spatial variations in power system inertia, it is necessary to conduct real-time assessment of the time-varying inertia of wind farms.

[0003] Existing online methods for assessing the time-varying inertia of wind farms are effective for identifying the constant inertia of synchronous motors, primarily traditional thermal power units. However, they fail to characterize the geographic and parameter variations within a wind farm, reducing the accuracy of their time-varying inertia assessment. Furthermore, limited observation conditions within wind farms are a significant constraint. The upload cycle of SCADA systems often fails to meet real-time requirements.

[0004] As the frequency components of wind turbines become increasingly complex, the demand for high-precision distributed monitoring equipment is becoming increasingly urgent. Therefore, how to accurately estimate the time-varying inertia of wind farms using limited information is an urgent problem that needs to be solved. Summary of the Invention

[0005] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a method and device for obtaining the time-varying inertia of a wind farm station based on a cluster aggregation method, the purpose of which is to solve the technical problem of how to accurately estimate the time-varying inertia of a wind farm using limited information.

[0006] To achieve the above objectives, according to one aspect of the present invention, a method for obtaining time-varying inertia of a wind farm based on a clustering aggregation method is provided, comprising:

[0007] S1: When a power imbalance event occurs, the frequency f of any fan in each unit is collected during the current sampling period. s and reactive power Qs ; The current sampling period is at least the third sampling period;

[0008] S2: The frequency f of any fan in the current sampling period s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power

[0009] S3: The equivalent frequency of each unit in the current sampling period and the two previous sampling periods and equivalent power are converted into snapshot matrix data, and a linear operator is calculated using every two consecutive snapshot matrix data to obtain two linear operators for each unit in the current sampling period;

[0010] S4: reconstructing the two linear operators corresponding to each unit in the current sampling period into the dynamic form of the rotor motion equation;

[0011] S5: solving the dynamic form of the rotor motion equation to obtain the equivalent inertia of each unit corresponding to the current sampling period;

[0012] S6: Perform weighted aggregation on the equivalent inertia corresponding to each unit in the current sampling period using capacity as weight to obtain the station inertia corresponding to the current sampling period.

[0013] In one embodiment, the S2 includes:

[0014] S21: Using the equivalent grid impedance Z of each unit eq , the frequency f collected by each unit in the current sampling period s and reactive power P s Construct the state estimation equation and measurement equation of the wind turbine;

[0015] S22: updating the state variables of the state estimation equation based on the Kalman filter structure;

[0016] S23: Correct the measurement equation using the updated state variable to obtain a corrected measurement equation and its posterior state variables Then the equivalent frequency corresponding to the current sampling period is obtained and equivalent power

[0017] in, is the a posteriori estimate of the unit terminal voltage amplitude, is the a posteriori estimate of the unit terminal voltage phase angle, is the prior state variable of the kth step, K k is the gain coefficient of the kth step, z k For real-time measurement, is the estimated amount of the measurement at step k.

[0018] In one embodiment, the S21 includes:

[0019] Using the formula Construct the state estimation equation; x k，1 、x k，2 、x k，3 and x k，4 They are the four state quantities of step k; f s，k-1 、P s，k-1 and Q s，k-1 are the equivalent frequency, equivalent active power and equivalent reactive power information estimated by the unit in the k-1th step; V s，k-1 and δ k-1 is the terminal voltage and phase angle estimated in the k-1th step; V t and θ are the voltage amplitude and phase angle at the common connection point; Z eq and φ are the magnitude and phase angle of the equivalent grid-connected impedance of the unit; ε1, ε2, ε3, and ε4 are the elements of the covariance matrix of the state estimation equation;

[0020] Using the formula Construct the measurement equation; f s，k 、V s，k , δ k and P s，k are the equivalent frequency, terminal voltage amplitude, terminal voltage phase angle and equivalent active power information estimated by the wind turbine in the kth step; V t and θ are the voltage amplitude and phase angle at the common connection point; Z eq and φ are the amplitude and phase angle of the equivalent grid-connected impedance of the unit; υ1, υ2 and υ3 are the elements of the covariance matrix of the measurement equation.

[0021] In one embodiment, the S22 includes:

[0022] Using the formula The state estimation equation and the measurement equation are updated; wherein, and are the estimated values ​​of the i-th state at the k-1th and k-th steps respectively, is the posterior state estimate at step k-1; is the prior state estimate at step k; n is the number of sigma random samples at each step; equations f(x) and h(x) represent the state estimation equations and measurement equations, respectively; is the prior covariance matrix of the k-th step; is the posterior covariance matrix of the k-1th step; u k is the input of step k; ε k and υ k is the error covariance matrix and noise covariance matrix of the kth step; is the i-th measurement value at the k-th step, is the average of all measurements at step k.

[0023] In one embodiment, before S21, the method further includes:

[0024] S01: Collect quasi-steady-state information of each unit in the wind farm, including capacity S j , current vector and voltage vector

[0025] S02: Perform equivalent aggregation on the quasi-steady-state information of the units in the same unit according to the aggregation formula, and calculate the equivalent grid-connected impedance Z of each unit eq .

[0026] In one embodiment, the S02 includes:

[0027] Using the formula Calculate the equivalent voltage and equivalent current of any unit A after aggregation;

[0028] Using the formula Calculate the equivalent impedance of any unit A;

[0029] Among them, j∈A represents all wind turbines in unit A, S eq_A is the equivalent capacity, and They are the unit equivalent current and the unit equivalent voltage respectively.

[0030] In one embodiment, the S3 includes:

[0031] Using the three sampling period unit power data P s and frequency data f s To construct the snapshot matrix X = [Δf s ΔP s ] T =[x1x2…x N ]∈R 2×N ; Where N is the total number of sequences of elements in the snapshot matrix, x1, x2 and x N Represent the data of the 1st, 2nd and Nth sequences in the snapshot matrix respectively;

[0032] Extract three consecutive sub-snapshot matrices from the snapshot matrix X, namely X N-2 =[x1x2…xN-2 ], X N-1 =[x2x3…x N-1 ] and X N =[x3x4…x N ];

[0033] X N-2 and X N-1 Substitute into the churn dynamic mode decomposition algorithm, update the relevant projection state matrix after orthogonalization, and obtain the Koopman linear operator corresponding to the previous sampling period t-1;

[0034] X N-1 and X N Substitute the churn dynamic mode decomposition algorithm to obtain the Koopman linear operator of the current sampling period t.

[0035] According to another aspect of the present invention, a device for obtaining time-varying inertia of a wind farm station based on a clustering aggregation method is provided, comprising:

[0036] The sampling module is used to collect the frequency f of any fan in each unit in the current sampling period when a power imbalance event occurs. s and reactive power Q s ; The current sampling period is at least the third sampling period;

[0037] The derivation module is used to convert the frequency f of any fan in the current sampling period into s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power

[0038] The calculation module is used to calculate the equivalent frequency of each unit in the current sampling period and the two previous sampling periods. and equivalent power are converted into snapshot matrix data, and a linear operator is calculated using every two consecutive snapshot matrix data to obtain two linear operators for each unit in the current sampling period;

[0039] A reconstruction module, configured to reconstruct the two linear operators corresponding to each unit in the current sampling period into a dynamic form of the rotor motion equation;

[0040] A solution module, configured to solve the dynamic form of the rotor motion equation to obtain the equivalent inertia of each unit corresponding to the current sampling period;

[0041] The aggregation module is used to perform weighted aggregation on the equivalent inertia corresponding to each unit in the current sampling period using capacity as a weight to obtain the station inertia corresponding to the current sampling period.

[0042] According to another aspect of the present invention, a control system for a wind farm is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

[0043] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are implemented.

[0044] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0045] (1) The present invention provides a method for obtaining the time-varying inertia of a wind farm station based on a clustering aggregation method, wherein the frequency f of any wind turbine in the current sampling period is s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power Then the equivalent frequency of each unit in the current sampling period and the two previous sampling periods is calculated. and equivalent power The inertia of each wind farm is converted into two linear operators, and the equivalent inertia of each wind farm during the current sampling period is calculated using the linear operators corresponding to each wind farm unit. Finally, a weighted calculation is used to determine the inertia level of the entire wind farm. This approach takes into account the differences among wind farm units, clustering and aggregating wind farm units with similar inertial responses, estimating the inertia of each aggregated unit separately, and then performing a weighted calculation to determine the inertia level of the entire wind farm. This approach has low computational complexity, accurately reflects the influence of parameter distribution, and estimates the time-varying inertia of the wind farm.

[0046] (2) This scheme uses the equivalent grid impedance Z of each unit eq , the frequency f collected by each unit in the current sampling period s and reactive power Q s Constructing the state estimation equation and measurement equation of the wind turbine; updating the state variables of the state estimation equation based on the Kalman filter structure; this method realizes the completion of the overall state estimation of the unit based on the minimum information requirement by considering the above information collected from any wind turbine in each unit.

[0047] (3) This scheme uses The state estimation equation is constructed; this method takes into account the uncertainty effect brought by sampling and realizes prior state estimation and noise filtering at the same time.

[0048] (4) This scheme uses the formula The state variables in the state estimation equation are updated and the measurement equation is corrected; this method combines real-time measurement with a priori state estimation result to correct the a priori state estimation result and obtain a posterior state estimation result.

[0049] (5) This scheme performs equivalent aggregation on the quasi-steady-state information of the units in the same unit according to the aggregation formula, and calculates the equivalent grid-connected impedance Z of each unit. eq This method takes into account the preliminary collection of unit steady-state information before the evaluation process, in order to prepare for the next step of calculating the unit grid-connected impedance information.

[0050] (6) This scheme uses the formula Calculate the equivalent impedance of any unit A; this method considers the use of the collected steady-state information to calculate the unit grid-connected impedance information.

[0051] (7) This scheme uses the power data of the unit P in three sampling periods s and frequency data f s A snapshot matrix is ​​constructed, and three consecutive sub-snapshot matrices are extracted from the snapshot matrix X. Then, two Koopman linear operators are calculated using the three sub-snapshot matrices. This method obtains the time-varying inertia of the unit in real time by obtaining the modal power and frequency data of each unit and combining it with the mechanism of the rotor motion equation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flow chart of a method for obtaining time-varying inertia of a wind farm based on a clustering aggregation method provided in Example 1 of the present invention;

[0053] Figure 2 This is a flow chart of another method for obtaining time-varying inertia of a wind farm based on a clustering aggregation method provided in Example 1 of the present invention;

[0054] Figure 3 A topological structure diagram of the IEEE39 node system including a wind farm provided in Example 1 of the present invention;

[0055] Figure 4 This is a graph showing the time-varying inertia assessment results of a wind farm provided in Scenario 1 of Example 1 of the present invention;

[0056] Figure 5 This is a curve chart of the time-varying inertia assessment results of a wind farm provided in Scenario 2 of Example 1 of the present invention. DETAILED DESCRIPTION

[0057] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0058] Example 1

[0059] like Figure 1 As shown, this embodiment provides a method for obtaining time-varying inertia of a wind farm based on a clustering aggregation method, including: S1-S6.

[0060] S1: When a power imbalance event occurs, the frequency f of any fan in each unit is collected during the current sampling period. s and reactive power Q s ; Among them, the current sampling period is at least the third sampling period.

[0061] S2: The frequency f of any fan in the current sampling period s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power

[0062] S3: The equivalent frequency of each unit in the current sampling period and the two previous sampling periods and equivalent power They are all converted into snapshot matrix data, and a linear operator is calculated using every two consecutive snapshot matrix data to obtain two linear operators for each unit in the current sampling period.

[0063] S4: Reconstruct the two linear operators corresponding to each unit in the current sampling period into the dynamic form of the rotor motion equation.

[0064] S5: Solve the dynamic form of the rotor motion equation to obtain the equivalent inertia of each unit corresponding to the current sampling period.

[0065] S6: The equivalent inertia corresponding to each unit in the current sampling period is weighted and aggregated with capacity as the weight to obtain the station inertia corresponding to the current sampling period.

[0066] like Figure 2 As shown in FIG, a wind farm station time-varying inertia assessment framework based on cluster aggregation method includes the following steps:

[0067] A. Determine the unit grouping index within the wind farm as the virtual inertia control coefficient;

[0068] B. Data acquisition and monitoring control system SCADA collects quasi-steady-state information of each unit in the wind farm, including capacity S j , current vector and voltage vector

[0069] C. Perform equivalent aggregation of the units of the same unit according to the aggregation formula and calculate the equivalent grid-connected impedance Z of each unit. eq ;

[0070] D. During the transient process after the power imbalance event occurs, the frequency of any fan in each unit is collected according to the sampling period. and reactive power

[0071] E. Based on the unscented Kalman filter structure, the equivalent frequency f of the unit is executed once in each sampling period. S and equivalent power P S Derivation of

[0072] F. Obtain the equivalent frequency f from the previous step S and equivalent power P S , using the snapshot matrix data of three sampling periods, the Koopman linear operator is obtained, thus completing the operator extraction of two sampling periods;

[0073] G. Combining the time domain reconstruction formula and the rotor motion equation, the linear operators of the two sampling periods are reconstructed into the dynamic equation relationship;

[0074] H. Solve the above equation to obtain the equivalent inertia of the unit in the current sampling period;

[0075] I. The equivalent inertia of each unit is weighted and aggregated using capacity as weight to obtain the station-level inertia level;

[0076] J. Continue the above process from D to I, and continue to update and calculate the time-varying inertia of the wind farm at the next moment.

[0077] The target system must meet the following conditions at the same time:

[0078] 1) A wide-area measurement system such as a phasor measurement unit (PMU) needs to be installed at the common connection point of the object being evaluated;

[0079] 2) The internal unit grid connection points of the evaluated object also need to be equipped with a wide-area measurement system such as the new energy equipment synchronous measurement device SMD-R;

[0080] 3) Collect relevant information for inertia assessment during the inertial response period after the power system suffers a large disturbance;

[0081] 4) The object under evaluation does not experience any unexpected situations such as disconnection or disconnection from the grid during the inertial response period.

[0082] As an optional implementation, S2 includes: S21: using the equivalent grid impedance Z of each unit eq , the frequency f collected by each unit in the current sampling period s and reactive power P s Construct the state estimation equation and measurement equation of the wind turbine; S22: Update the state variables of the state estimation equation based on the Kalman filter structure; S23: Use the updated state variables to correct the measurement equation to obtain the corrected measurement equation and its posterior state variables Then get the equivalent frequency corresponding to the current sampling period and equivalent power in, is the a posteriori estimate of the unit terminal voltage amplitude, is the a posteriori estimate of the unit terminal voltage phase angle, is the prior state variable of the kth step, K k is the gain coefficient of the kth step, z k For real-time measurement, is the estimated amount of the measurement at step k.

[0083] As an optional implementation, S21 includes: using the formula Construct the state estimation equation; x k，1 、x k，2 、x k，3 and x k，4 They are the four state quantities of step k; f s，k-1 、P s，k-1 and Q s，k-1 are the equivalent frequency, equivalent active power and equivalent reactive power information estimated by the unit in the k-1th step; V s，k-1 and δ k-1 is the terminal voltage and phase angle estimated in the k-1th step; V t and θ are the voltage amplitude and phase angle at the common connection point; Z eq and φ are the amplitude and phase angle of the equivalent grid-connected impedance of the unit; ε1, ε2, ε3 and ε4 are the elements of the covariance matrix of the state estimation equation.

[0084] Using the formula Construct the measurement equation; f s，k 、V s，k , δ k and P s，k are the equivalent frequency, terminal voltage amplitude, terminal voltage phase angle and equivalent active power information estimated by the wind turbine in the kth step; V tand θ are the voltage amplitude and phase angle at the common connection point; Z eq and φ are the amplitude and phase angle of the equivalent grid-connected impedance of the unit; υ1, υ2 and υ3 are the elements of the covariance matrix of the measurement equation.

[0085] As an optional implementation, S22 includes: using the formula Update the state estimation equation and measurement equation; where, and are the estimated values ​​of the i-th state at the k-1th and k-th steps respectively, is the posterior state estimate at step k-1; is the prior state estimate at step k; n is the number of sigma random samples at each step; equations f(x) and h(x) represent the state estimation equations and measurement equations, respectively; is the prior covariance matrix of the k-th step; is the posterior covariance matrix of the k-1th step; u k is the input of step k; ε k and υ k is the error covariance matrix and noise covariance matrix of the kth step; is the i-th measurement value at the k-th step, is the average of all measurements at step k.

[0086] Specifically, first according to the structure of the Kalman filter:

[0087]

[0088] where x k =f(x k-1 ,u k )+ε k is the state estimation equation, z k =h(x k ,u k )+υ k is the measurement equation.

[0089] Considering the power system flow calculation, the following formula is valid:

[0090]

[0091] Therefore, the following state estimation equations can be written:

[0092]

[0093] Where k represents the number of steps, f s，k 、P s，k and Q s，k are the equivalent frequency, equivalent active power and equivalent reactive power information estimated by the unit in the previous step; Vs，k and δ k is the terminal voltage and phase angle estimated in the previous step of the measured wind turbine; V t and θ are the voltage amplitude and phase angle at the common connection point; Z and φ are the amplitude and phase angle of the equivalent grid impedance of the unit. ε1, ε2, ε3, and ε4 are the elements of the covariance matrix of the state estimation equation.

[0094] At the same time, the device-level PMU deployed in the wind farm can help collect the terminal frequency f of a single wind turbine. s , and approximate it to the sampling equivalent frequency of the entire cluster; the PMU deployed at the common connection point can measure the output active power of the common connection point, which is equivalent to the sum of the equivalent output power of all units ∑P s，k In addition, the reactive power Q of each wind turbine is s remains almost unchanged, that is, its reference value Q sref Therefore, the measured variable is z = [Q sref , f s ,∑P s ] T The following measurement equations can be written:

[0095]

[0096] Taking the kth step as an example, the unscented Kalman filter first completes the prediction step through the following formula:

[0097]

[0098] Then complete the update step through the following formula:

[0099]

[0100] Where n is the number of sampling points, and is the covariance matrix of the state estimation equation and the measurement equation. The posterior state estimate state variable Contains the power and frequency information of each unit, which will be used in the next section.

[0101] As an optional implementation, before S21, the following steps are included: S01: collecting quasi-steady-state information of each unit in the wind farm, including capacity S j , current vector and voltage vector S02: Perform equivalent aggregation of the quasi-steady-state information of the units in the same unit according to the aggregation formula, and calculate the equivalent grid-connected impedance Z of each unit eq .

[0102] The inertia response differences between wind turbines at a wind farm are influenced by a variety of factors, primarily categorized into two categories: operating conditions and model parameter factors. Operating conditions include wind speed and distance from the common connection point (i.e., layout). Model parameter factors include the virtual inertia control parameter, droop control gain, virtual inertia response delay, and the rotor's inherent rotational inertia. For site-level inertia assessment, the dominant parameter is identified as the virtual inertia control parameter, which can then be used as a unit grouping indicator.

[0103] The data acquisition and monitoring control system SCADA collects the quasi-steady-state information of each unit in the wind farm, including capacity S j , current vector and voltage vector The SCADA system collects information on the capacity, current amplitude, and voltage amplitude of each unit and uses state estimation techniques to infer phase angles. The upload cycle for the SCADA system is typically 1-2 minutes, which meets the current needs for quasi-steady-state information collection.

[0104] As an optional implementation, S02 includes: performing equivalent aggregation on the units of the same unit according to the aggregation formula, and calculating the equivalent grid-connected impedance Z of each unit. eq SCADA collects quasi-steady-state information of each unit in the wind farm, including capacity S j , machine-end current vector and the terminal voltage vector Where subscript j is the number of the wind turbine. Taking a certain unit A as an example, the calculation method of its equivalent voltage and equivalent current after aggregation is:

[0105]

[0106] Among them, j∈A represents all wind turbines in unit A, S eq_A is the equivalent capacity, and They are the unit equivalent current and the unit equivalent voltage respectively.

[0107] Using the above-mentioned equivalent current of the unit and the equivalent voltage of the unit Can be used for calculation. The corresponding relationship is: in, is the equivalent grid-connected impedance of the unit obtained, is the voltage at the common connection point.

[0108] Regarding the flow-based dynamic modal decomposition method, based on the power and frequency information of each unit obtained in the previous step, the time-varying inertia of each unit is calculated and aggregated into the station-level inertia. The specific steps are as follows:

[0109] As an optional implementation, S3 includes: using the power data P of the unit in three sampling periods s and frequency data f s To construct the snapshot matrix X = [Δf s ΔP s ] T =[x1x2…x N ]∈R 2×N ; Where N is the total number of sequences of elements in the snapshot matrix, x1, x2 and x N Represent the data of the 1st, 2nd and Nth sequences in the snapshot matrix respectively; extract three consecutive sub-snapshot matrices from the snapshot matrix X, namely X N-2 =[x1x2…x N-2 ], X N-1 =[x2x3…x N-1 ] and X N =[x3x4…x N ]; X N-2 and X N-1 Substitute into the SDMD loss dynamic mode decomposition algorithm, after Gram-Schmidt orthogonalization, update the relevant projection state matrix, obtain the linear operator corresponding to the sampling period t-1, and then convert X N-1 and X N Substitute into the SDMD algorithm to obtain the linear operator corresponding to the sampling period t.

[0110] Furthermore, combined with the rotor motion equation, based on the time domain reconstruction formula of the linear operator proposed in the previous section, two equations corresponding to the t-1 sampling period and the t sampling period are formed as follows:

[0111] The dynamic form of the rotor motion equation for the t-1 sampling period is:

[0112]

[0113] The dynamic form of the rotor motion equation for the sampling period t is:

[0114]

[0115] In matrix form, the characteristic root factor e λt It is in the form of a diagonal matrix, a full-rank matrix, and can be simplified.

[0116] After simplification, the original system of equations can be written in the form of the following matrix:

[0117]

[0118] The time-varying inertia H of each unit obtained in the above formula is v, and perform the following weighted calculation with capacity as weight: Among them S i is the rated capacity of each unit, H WF is the final time-varying inertia of the wind farm. At the next moment, the frequency and reactive power of any wind turbine in each unit are collected and the above steps are repeated.

[0119] The effects of the embodiments are shown below to illustrate the effectiveness of the present invention.

[0120] Scenario 1: Identical wind turbines are arranged neatly.

[0121] exist Figure 3 The topology of the IEEE 39-node system with a wind farm shown in the figure includes the IEEE standard 39-node system and an additional wind farm with 25 wind turbines. G1, G2, G3, G4, G5, G6, G7, G8, G9, and G10 are traditional synchronous generators; the wind farm is integrated into node 5; and node 39 is set as a balancing node.

[0122] At sampling period t=0s, the load of node 8 is set to increase suddenly by 10%, thereby triggering the inertia response of the wind farm. The same wind turbines are arranged in a cluster, and the specific parameters are shown in Table 1.

[0123] Table 1

[0124]

[0125]

[0126] The time-varying inertia of the wind farm evaluated by the proposed technical framework is as follows: Figure 4 shown.

[0127] During the startup phase, the estimated results maintain a certain distance from the theoretical true values. This is because the unscented Kalman filter requires a certain number of iterations to accurately estimate the true power and frequency information for each turbine. Furthermore, the gap between the estimated and true values ​​decreases over time. During the initial period of a large disturbance, the estimated results quickly track the theoretical true values ​​of the wind farm's time-varying inertia, demonstrating the effectiveness of the proposed technical framework for online evaluation of wind farm time-varying inertia.

[0128] Scenario 2: The same wind turbines are completely hashed.

[0129] exist Figure 3 In the IEEE39 node system with wind farms shown in FIG, wind turbines with the same settings are hashed and distributed, with specific parameters shown in Table 2.

[0130]

[0131]

[0132] and Figure 4 compared to, Figure 5 The time-varying inertia assessment results fluctuate more significantly. This is due to the greater variability in operating conditions among wind turbines within the same unit, which reduces the accuracy of the aggregation. However, the proposed technical framework maintains a maximum relative error of no more than 5% and an average error of no more than 3%, still meeting the requirements for online assessment of time-varying inertia at wind farms.

[0133] Example 2

[0134] This embodiment provides a wind farm station time-varying inertia acquisition device based on the clustering aggregation method, including: a sampling module, a derivation module, a calculation module, a reconstruction module, a solution module, and an aggregation module. The sampling module is used to collect the frequency f of any wind turbine in each unit in the current sampling period when a power imbalance event occurs. s and reactive power Q s The current sampling period is at least the third sampling period. The derivation module is used to convert the frequency f of any fan in the current sampling period into s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power Calculation module, used to calculate the equivalent frequency of each unit in the current sampling period and the two previous sampling periods and equivalent power All of them are converted into snapshot matrix data. A linear operator is calculated using every two consecutive snapshot matrix data to obtain two linear operators for each unit in the current sampling period. The reconstruction module is used to reconstruct the two linear operators corresponding to each unit in the current sampling period into the dynamic form of the rotor motion equation. The solution module is used to solve the dynamic form of the rotor motion equation to obtain the equivalent inertia corresponding to each unit in the current sampling period. The aggregation module is used to perform weighted aggregation on the equivalent inertia corresponding to each unit in the current sampling period using capacity as the weight to obtain the station inertia corresponding to the current sampling period.

[0135] Example 3

[0136] This embodiment provides a control system for a wind farm station, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

[0137] Example 4

[0138] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method are implemented.

[0139] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for obtaining time-varying inertia of a wind farm based on clustering and aggregation method, characterized in that: include: S1: When a power imbalance event occurs, the frequency f of any fan in each unit is collected during the current sampling period. s and reactive power Q s ; The current sampling period is at least the third sampling period; S2: The frequency f of any fan in the current sampling period s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power S3: The equivalent frequency of each unit in the current sampling period and the two previous sampling periods and equivalent power are all converted into snapshot matrix data, and a linear operator is calculated using every two consecutive snapshot matrix data to obtain two linear operators for each unit in the current sampling period; S4: reconstructing the two linear operators corresponding to each unit in the current sampling period into the dynamic form of the rotor motion equation; S5: solving the dynamic form of the rotor motion equation to obtain the equivalent inertia of each unit corresponding to the current sampling period; S6: Perform weighted aggregation on the equivalent inertia corresponding to each unit in the current sampling period using capacity as weight to obtain the station inertia corresponding to the current sampling period.

2. The method for obtaining time-varying inertia of a wind farm based on the clustering aggregation method according to claim 1, characterized in that: The S2 includes: S21: Using the equivalent grid impedance Z of each unit eq , the frequency f collected by each unit in the current sampling period s and reactive power P s Construct the state estimation equation and measurement equation of the wind turbine; S22: Based on the Kalman filter structure, the state variables of the state estimation equation are updated, and the measurement equation is corrected using the updated state variables to obtain a corrected measurement equation. and its posterior state variables Then the equivalent frequency corresponding to the current sampling period is obtained and equivalent power in, is the a posteriori estimate of the unit terminal voltage amplitude, is the a posteriori estimate of the unit terminal voltage phase angle, is the prior state variable of the kth step, K k is the gain coefficient of the kth step, z k For real-time measurement, is the estimated amount of the measurement at step k.

3. The method for obtaining time-varying inertia of a wind farm based on clustering and aggregation method according to claim 2, characterized in that: The S21 includes: Using the formula Establish the state estimation equation; k，1 、x k，2 、x k，3 and x k，4 They are the four state quantities of step k; f s，k-1 、P s，k-1 and Q s，k-1 are the equivalent frequency, equivalent active power and equivalent reactive power information estimated by the unit in the k-1th step; V s，k-1 and δ k-1 is the terminal voltage and phase angle estimated in the k-1th step; V t and θ are the voltage amplitude and phase angle at the common connection point; Z eq and φ are the magnitude and phase angle of the equivalent grid-connected impedance of the unit; ε1, ε2, ε3, and ε4 are the elements of the covariance matrix of the state estimation equation; Using the formula Construct the measurement equation; f s，k 、V s，k , δ k and P s，k are the equivalent frequency, terminal voltage amplitude, terminal voltage phase angle and equivalent active power information estimated by the wind turbine in the kth step; V t and θ are the voltage amplitude and phase angle at the common connection point; Z eq and φ are the amplitude and phase angle of the equivalent grid-connected impedance of the unit; υ1, υ2 and υ3 are the elements of the covariance matrix of the measurement equation.

4. The method for obtaining time-varying inertia of a wind farm based on cluster aggregation method according to claim 3, characterized in that: The S22 includes: Using the formula The state variables in the state estimation equation are updated and the measurement equation is corrected; wherein, and are the estimated values ​​of the i-th state at the k-1th and k-th steps respectively, is the posterior state estimate at step k-1; is the prior state estimate at step k; n is the number of sigma random samples at each step; equations f(x) and h(x) represent the state estimation equations and measurement equations, respectively; is the prior covariance matrix of the k-th step; is the posterior covariance matrix of the k-1th step; u k is the input of step k; ε k and v k is the error covariance matrix and noise covariance matrix of the kth step; is the i-th measurement value at the k-th step, is the average of all measurements at step k.

5. The method for obtaining time-varying inertia of a wind farm based on clustering and aggregation method according to claim 2, characterized in that: Before S21, the method further includes: S01: Collect quasi-steady-state information of each unit in the wind farm, including capacity S j , current vector and voltage vector S02: Perform equivalent aggregation on the quasi-steady-state information of the units in the same unit according to the aggregation formula, and calculate the equivalent grid-connected impedance Z of each unit eq .

6. The method for obtaining time-varying inertia of a wind farm based on cluster aggregation method according to claim 5, characterized in that: The S02 includes: Using the formula Calculate the equivalent voltage and equivalent current of any unit A after aggregation; Using the formula Calculate the equivalent impedance of any unit A; Among them, j∈A represents all wind turbines in unit A, S e q A is the equivalent capacity, and They are the unit equivalent current and the unit equivalent voltage respectively.

7. The method for obtaining time-varying inertia of a wind farm based on a clustering aggregation method according to any one of claims 1 to 6, characterized in that: The S3 includes: Using the three sampling period unit power data P s and frequency data f s To construct the snapshot matrix X = [Δf s ΔP s ] T =[x1x2…x N ]∈R 2×N ; Where N is the total number of sequences of elements in the snapshot matrix, x1, x2 and x N Represent the data of the 1st, 2nd and Nth sequences in the snapshot matrix respectively; Extract three consecutive sub-snapshot matrices from the snapshot matrix X, namely X N-2 =[x1x2…x N-2 ], X N-1 =[x2x3…x N-1 ] and X N =[x3x4…x N ]; X N-2 and X N-1 Substitute into the churn dynamic mode decomposition algorithm, update the relevant projection state matrix after orthogonalization, and obtain the Koopman linear operator corresponding to the previous sampling period t-1; X N-1 and X N Substitute the churn dynamic mode decomposition algorithm to obtain the Koopman linear operator of the current sampling period t.

8. A device for obtaining time-varying inertia of a wind farm station based on a clustering aggregation method, characterized in that: include: The sampling module is used to collect the frequency f of any fan in each unit in the current sampling period when a power imbalance event occurs. s and reactive power Q s ; The current sampling period is at least the third sampling period; The derivation module is used to convert the frequency f of any fan in the current sampling period into s and reactive power Q s Derived as the equivalent frequency of any fan unit and equivalent power The calculation module is used to calculate the equivalent frequency of each unit in the current sampling period and the two previous sampling periods. and equivalent power are all converted into snapshot matrix data, and a linear operator is calculated using every two consecutive snapshot matrix data to obtain two linear operators for each unit in the current sampling period; A reconstruction module, configured to reconstruct the two linear operators corresponding to each unit in the current sampling period into a dynamic form of the rotor motion equation; A solution module, configured to solve the dynamic form of the rotor motion equation to obtain the equivalent inertia of each unit corresponding to the current sampling period; The aggregation module is used to perform weighted aggregation on the equivalent inertia corresponding to each unit in the current sampling period using capacity as a weight to obtain the station inertia corresponding to the current sampling period.

9. A control system for a wind farm station, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images