Photovoltaic power station step-by-step clustering and equivalent modeling method based on improved k-means algorithm

Abstract

The invention belongs to the technical field of photovoltaic power generation and grid connection, and particularly provides a photovoltaic power station step-by-step clustering and equivalent modeling method based on an improved k-means algorithm, and the method comprises the following steps: 1, building a detailed simulation model of a photovoltaic power station, constructing an electrical topological structure of the photovoltaic power station, and inputting the electrical parameters of each module; 2, carrying out load flow calculation and transient calculation on the detailed simulation model of the photovoltaic power station, and collecting power and voltage values during a steady state and a transient state; step 3, clustering by adopting an improved k-means algorithm: carrying out primary clustering by taking the active power of the photovoltaic power station in the steady state period as an index, and then carrying out secondary clustering by taking the active power and reactive power of the photovoltaic power station in the transient state period as indexes; and 4, aggregating electrical parameters in the detailed simulation model of the photovoltaic power station according to a grouping result, and building an equivalent simulation model of the photovoltaic power station. According to the method, the multi-grouping index is introduced, and the accuracy of equivalent modeling is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic power generation and grid connection, and particularly relates to a method for step-by-step clustering and equivalent modeling of a photovoltaic power station based on an improved k-means algorithm. Background Art

[0002] In a photovoltaic power station, at least dozens or even hundreds of units are equipped, showing the characteristics of a complex topological structure, a large number of nodes, and frequent switching of power electronic switches. If the detailed model of the power station is directly simulated, not only is the workload huge, but also the hardware requirements for the test platform of the digital model of the operating power station are relatively high. Therefore, it is necessary to simplify the power station model, which can not only accurately reflect the grid connection performance of the power station, but also solve the problems of simulation scale and accuracy. The equivalent of the power station can be divided into single-machine and multi-machine equivalents. Aiming at the problem of large errors in the single-machine equivalent model, a multi-machine equivalent model is currently mostly used to equivalent the power station. And how to select a reasonable clustering index is the key problem to be considered in multi-machine equivalent modeling.

[0003] During the steady-state operation of a photovoltaic power station, it is affected by various factors, such as light intensity, temperature, geographical location, component aging degree, and the topological structure of the power station. The active output powers of different units are different, while the reactive power output is approximately 0. During the fault ride-through period, the voltage drop values of the units at different positions are different, resulting in differences in the dynamic support capabilities provided by each inverter. Therefore, it is difficult to comprehensively reflect these differences by using a single index for clustering, which is likely to introduce large equivalent errors, thus affecting the analysis of the overall characteristics of the power station and the evaluation of the fault ride-through performance. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for step-by-step clustering and equivalent modeling of a photovoltaic power station based on an improved k-means algorithm, so as to improve the accuracy of equivalent modeling.

[0005] To solve the above technical problem, the technical solution adopted by the present invention is: a method for step-by-step clustering and equivalent modeling of a photovoltaic power station based on an improved k-means algorithm, including the following steps: Step 1: Build a detailed simulation model of the photovoltaic power station, construct the electrical topological structure of the photovoltaic power station, and input the electrical parameters of each module; Step 2: Perform power flow calculation and transient calculation on the detailed simulation model of the photovoltaic power station, and collect the power and voltage values during the steady state and transient state; Step 3: Use the improved k-means algorithm for clustering: initially cluster with the active power during the steady state of the photovoltaic power station as the index, and then perform secondary clustering with the active and reactive powers during the transient state of the photovoltaic power station as the index; Step 4: Aggregate the electrical parameters in the detailed simulation model of the photovoltaic power station according to the clustering results, and build an equivalent simulation model of the photovoltaic power station.

[0006] In the preferred solution, in the first step, the active and reactive power control strategies during the low voltage ride-through of the photovoltaic generator sets in the detailed simulation model of the photovoltaic power station are as follows: Output active power during the steady state, and reactive power is 0; during the transient state, the adopted active and reactive power control strategies are as follows: The active power control strategy is as follows: (1); Wherein, K 1_Ip is the active current coefficient 1; K 2_Ip is the active current coefficient 2; Ip set is the active current coefficient 3; Ip LVRT is the active current during the fault; Ip 0 is the active current before the fault; Vt is the terminal voltage amplitude; The reactive power control strategy is as follows: (2); Wherein, K 1_Iq is the reactive current coefficient 1; K 2_Iq is the reactive current coefficient 2; Iq set is the reactive current coefficient 3; Iq LVRT is the reactive current during the fault; Iq 0 is the reactive current before the fault; VL in is the low voltage ride-through threshold; Vt is the terminal voltage amplitude.

[0007] In the preferred solution, in the second step, the Newton-Raphson method is used for the power flow calculation of the photovoltaic power station simulation model.

[0008] In the preferred solution, in the second step, the transient calculation of the photovoltaic power station simulation model is carried out: a three-phase fault is set at the grid connection point to make the grid connection point voltage enter the low voltage ride-through state, and the active and reactive powers at the machine terminal of the photovoltaic generator set are collected.

[0009] In the preferred solution, in the third step, the improved k-means algorithm is used to cluster the photovoltaic power station, and the specific operation is as follows: S301. Normalize the clustering metrics in the PV power station. The initial clustering metric is the active power during the steady state, and the metrics for the secondary clustering are the active and reactive powers during the transient state. Set the number of cluster clusters in the PV power station to k , and randomly initialize k cluster centers: (3); In the formula, is the set of initial values of the k -th cluster center; represents the initial value of the k -th cluster center on the i -th feature dimension; S302. Assign samples to clusters. Assign each sample point x i to the nearest cluster center. The weighted Euclidean distance calculation formula from x i to each cluster center is: (4); Among them, x ij is the x i -th eigenvalue of the sample j ; is the k -th eigenvalue of the center point of cluster j ; is the weight of feature j , indicating the importance of this feature for distance calculation; S303. Update the cluster centers. According to the assignment results, calculate the new center of each cluster: (5); In the formula, is the value of the new center of the k -th cluster on the j -th feature; C k is the sample set of the k -th cluster; Compare the change amount of the cluster center before and after the update. If this change amount is less than the preset threshold, it is considered that the cluster center has converged; S304. Calculate the silhouette coefficient. For a sample x i , the silhouette coefficient is defined as: (6); Among them, a ([[]] i ) is the sample xi The average distance to other samples in its affiliated cluster, with the expression: (7); In the formula, | μ k | is the number of samples in the cluster μ k ; is the sample x i and x j The Euclidean distance between; b ( i ) is the average distance of the sample x i to the nearest other cluster sample, with the expression: (8); In the formula, is the number of samples in the cluster ; l ≠ k indicates that the cluster l is other clusters except the cluster to which x i belongs; S305. Calculate the overall silhouette coefficient. The overall silhouette coefficient of the clustering result is the average of the silhouette coefficients of all samples: (9); During each iteration, record the randomly selected k and w and the corresponding overall silhouette coefficient S ([[]] k , w ). After multiple iterations, take S ([[]] k , w ) corresponding to the maximum k and w .

[0010] In the preferred solution, in step four, aggregate the electrical parameters in the photovoltaic power station according to the clustering result. The specific aggregation formula is as follows: S401. The equivalent unit parameters in the photovoltaic power station are shown in the following formula: (10); In the formula, S eq , P eq and Q eqThey are the equivalent unit capacity, active power, and reactive power of the units within the group in the PV power station, respectively; S i 、 P i and Q i They are the capacity, active power, and reactive power of the i th unit within the group, respectively; S402. The parameters of the equivalent transformer substation in the PV power station are shown in the following formula: (11); In the formula: S Teq is the capacity of the equivalent transformer substation; S Ti is the capacity of the i th transformer substation; Z Teq is the impedance of the equivalent transformer substation; δ i is the weight of the capacity of the i th transformer substation in the total capacity of the transformer substations; Z Ti is the impedance of the i th transformer substation; S403. In the PV power station, the topologies of the collector lines are divided into two types, namely the trunk type structure and the radial type structure; When connected in the trunk type, the equivalent impedance of the collector line is: (12); In the formula, Z eq is the equivalent impedance of the collector line, Z 1 ~ Z N are the collector line impedances from the 1st unit to the Nth unit to the bus of the feeder connection point; When connected in the radial type, the equivalent impedance of the collector line is: (13); In the formula, Z i is the collector line impedance from the i th unit to the bus of the feeder connection point; S404. According to the clustering results, the photovoltaic units in the same group are regarded as an equivalent unit, and the corresponding equivalent transformer and collection line models are built: first, the capacity, active power and reactive power of the units in the cluster are extracted, and the equivalent units are aggregated using the formula of S401; then, the transformer parameters are aggregated, and the aggregation method of S402 is used to ensure that the equivalent transformer capacity, voltage level and impedance are consistent with the corresponding detailed model; then, the collection line is equivalently simplified, and the equivalent impedance of the trunk and radial structures is calculated according to the aggregation formula of S403; finally, all the calculation results are filled into the corresponding module to complete the construction of the equivalent model.

[0011] The present invention provides a photovoltaic power station step-by-step clustering and equivalent modeling method based on an improved k-means algorithm, which has the following beneficial effects: 1. The method of the present invention significantly improves the accuracy and efficiency of photovoltaic power station clustering by introducing multiple clustering indicators, including active power during steady state and active and reactive power during transient period, and combines the step-by-step clustering strategy, thereby enhancing the applicability of the equivalent model in dynamic scenarios.

[0012] 2. The present invention uses the k-means algorithm to group the units in the photovoltaic power station and dynamically adjusts the number of clusters in combination with the silhouette coefficient. k and indicator weights w , reducing human intervention, balancing the closeness and separation of clustering results, and further improving the efficiency and accuracy of equivalent modeling.

[0013] 3. Step 1 of the present invention provides comprehensive and accurate basic data support for clustering and equivalent modeling by building a detailed simulation model of the photovoltaic power station, constructing the electrical topology and entering the electrical parameters of each module. The detailed simulation model can accurately reflect the electrical topology and component characteristics of the system, ensuring the reliability of subsequent calculation results.

[0014] 4. Step 2 of the present invention combines power flow calculation with transient simulation to collect power and voltage values ​​during steady state and transient state, fully reflecting the static characteristics and dynamic response characteristics of the photovoltaic power station. Power flow calculation provides active power distribution under steady-state operation, while transient simulation captures the dynamic behavior of the system under disturbance conditions. The combination of the two constitutes a multi-dimensional clustering index system. This method effectively makes up for the limitation that traditional clustering is based only on single static data, and provides reliable guarantee for the applicability of the model in dynamic scenarios.

[0015] 5. Step 3 of the present invention uses an improved k-means algorithm for clustering, and achieves accurate and efficient clustering through a step-by-step clustering strategy. The initial clustering uses steady-state active power as an indicator to quickly screen out units with similar characteristics; the secondary clustering introduces active and reactive power during transient periods to further optimize the clustering results. At the same time, the number of clusters is dynamically adjusted in combination with the silhouette coefficient.k and index weights w Automatically optimize the clustering parameters to reduce the influence of human intervention on the results. This clustering method effectively improves the accuracy and efficiency of photovoltaic power station clustering, and significantly enhances the scientificity and practicality of the clustering strategy.

[0016] 6. In step four of the present invention, the electrical parameters of the photovoltaic power station are aggregated according to the clustering results, and an equivalent simulation model is built, which greatly improves the modeling efficiency and ensures the simulation accuracy under dynamic scenarios. By aggregating the unit parameters after clustering into an equivalent model, the modeling process of complex systems is significantly simplified, and the simulation calculation amount is reduced. At the same time, the clustering is based on multi-dimensional steady-state and transient indicators, enabling the equivalent model to accurately reflect the dynamic response characteristics of the photovoltaic power station, providing an efficient and accurate modeling tool for power grid simulation, planning, and operation. Brief Description of the Drawings

[0017] The present invention will be further described below in conjunction with the drawings and embodiments: Figure 1 is the flow chart of the step-by-step clustering and equivalence of the photovoltaic power station based on the improved k-means algorithm of the present invention; Figure 2 is the voltage waveform comparison diagram of the detailed simulation model and the equivalent simulation model of the photovoltaic power station in the present invention; Figure 3 is the current waveform comparison diagram of the detailed simulation model and the equivalent simulation model of the photovoltaic power station in the present invention; Figure 4 is the active power waveform comparison diagram of the detailed simulation model and the equivalent simulation model of the photovoltaic power station in the present invention; Figure 5 is the reactive power waveform comparison diagram of the detailed simulation model and the equivalent simulation model of the photovoltaic power station in the present invention. Detailed Embodiment

[0018] As Figure 1 shown, a step-by-step clustering and equivalence modeling method for a photovoltaic power station based on an improved k-means algorithm includes the following steps: Step 1: Build a detailed simulation model of the photovoltaic power station, construct the electrical topology structure of the photovoltaic power station, including power generation units, collector lines, outgoing lines, box transformers, main transformers, etc., and input the electrical parameters of each module, including the power values of photovoltaic generator sets, impedance parameters of box transformers, main transformers, and collector lines.

[0019] The active and reactive power control strategies during the low voltage ride-through period of the photovoltaic generator set in the detailed simulation model of the photovoltaic power station are as follows: Output active power during the steady state period, and the reactive power is 0; during the transient period, the adopted active and reactive power control strategies are as follows: The active power control strategy is as follows: (1); Wherein, K 1_Ip is the active current coefficient 1; K 2_Ip is the active current coefficient 2; Ip set is the active current coefficient 3; Ip LVRT is the active current during the fault; Ip 0 is the active current before the fault; Vt is the terminal voltage amplitude; The reactive power control strategy is as follows: (2); Wherein, K 1_Iq is the reactive current coefficient 1; K 2_Iq is the reactive current coefficient 2; Iq set is the reactive current coefficient 3; Iq LVRT is the reactive current during the fault; Iq 0 is the reactive current before the fault; VL in is the low voltage ride-through threshold; Vt is the terminal voltage amplitude.

[0020] Step 2: Perform power flow calculation and transient calculation on the detailed simulation model of the PV power station, and collect the power and voltage values during steady state and transient state.

[0021] The Newton-Raphson method, i.e., the Newton-Raphson method, is used for power flow calculation of the PV power station simulation model.

[0022] Perform transient calculation on the PV power station simulation model: Set a three-phase fault at the grid connection point to make the grid connection point voltage enter the low voltage ride-through state, and collect the active and reactive powers at the machine terminal of the PV generator set.

[0023] Step 3: Use the improved k-means algorithm for clustering: Initially cluster with the active power during the steady state of the PV power station as the index, and then perform secondary clustering with the active and reactive powers during the transient state of the PV power station as the index.

[0024] Use the improved k-means algorithm to cluster the PV power station. The specific operations are as follows: S301. Normalize the clustering index in the PV power station. The initial clustering index is the active power during the steady state, and the index for secondary clustering is the active and reactive powers during the transient state. Set the number of clusters for the units in the PV power station tok , and randomly initialize k cluster centers: (3); In the formula, is the initial value of the k th cluster center; represents the initial value of the k th cluster center on the i th feature dimension; S302. Perform cluster assignment on the samples, and assign each sample point x i to the nearest cluster center. x i The weighted Euclidean distance calculation formula from each sample to each cluster center is: (4); Among them, x ij is the x i th eigenvalue of the sample j ; is the k th eigenvalue of the center point of cluster j ; is the weight of feature j , indicating the importance of this feature for distance calculation; S303. Update the cluster centers. According to the assignment results, calculate the new center of each cluster: (5); In the formula, is the value of the new center of the k th cluster on the j th feature; C k is the sample set of the k th cluster; Compare the change amount of the cluster center before and after the update. If this change amount is less than the preset threshold, it is considered that the cluster center has converged; S304. Calculate the silhouette coefficient. For a sample x i , the silhouette coefficient is defined as: (6); Among them, a ( i ) is the average distance from the sample x i to other samples in its affiliated cluster, and the expression is: (7); In the formula, |μ k | is the number of samples in the cluster μ k ; is the sample x i and x j the Euclidean distance between; b ( i ) is the average distance of the sample x i to the nearest other cluster sample, and the expression is: (8); In the formula, is the number of samples in the cluster ; l ≠ k indicates that the cluster l is other clusters except the cluster x i belongs to; S305. Calculate the overall silhouette coefficient. The overall silhouette coefficient of the clustering result is the average of the silhouette coefficients of all samples: (9); In each iteration process, record the randomly selected k and w and the corresponding overall silhouette coefficient S ( k , w ). After multiple iterations, take S ( k , w ) corresponding to the maximum k and w .

[0025] Step Four: Aggregate the electrical parameters in the detailed simulation model of the photovoltaic power station according to the clustering result, and build an equivalent simulation model of the photovoltaic power station.

[0026] Aggregate the electrical parameters in the photovoltaic power station according to the clustering result. The specific aggregation formula is as follows: S401. The equivalent unit parameters in the photovoltaic power station are shown in the following formula: (10); In the formula, S eq , P eq and Q eq are the equivalent unit capacity, active power, and reactive power within the group in the photovoltaic power station respectively; Si , P i and Q i are respectively the capacity, active power and reactive power of the i th unit in the group;

[0027] S402. The equivalent box transformer parameters in the PV power station are shown in the following formula: (11); In the formula: S Teq is the equivalent box transformer capacity; S Ti is the capacity of the i th box transformer; Z Teq is the equivalent box transformer impedance; δ i is the weight of the capacity of the i th box transformer in the total box transformer capacity; Z Ti is the impedance of the i th box transformer; S403. In the PV power station, the collector line topological structure is divided into two types, namely the trunk type structure and the radial type structure; When connected in the trunk type, the equivalent impedance of the collector line is: (12); In the formula, Z eq is the equivalent impedance of the collector line, Z 1 ~ Z N are the collector line impedances from the 1st unit to the Nth unit to the bus of the feeder connection point; When connected in the radial type, the equivalent impedance of the collector line is: (13); In the formula, Z i is the collector line impedance from the i th unit to the bus of the feeder connection point; S404. According to the clustering results, the photovoltaic units within the same group are regarded as an equivalent unit, and the corresponding equivalent substation and collector line models are built: First, extract the capacity, active power, and reactive power of the units within the cluster, and aggregate them into an equivalent unit using the formula in S401. Then, aggregate the substation parameters, and adopt the aggregation method in S402 to ensure that the capacity, voltage level, and impedance of the equivalent substation are consistent with the corresponding detailed model. Next, perform equivalent simplification on the collector line, and calculate the equivalent impedance of the main-line type and radial type structures according to the aggregation formula in S403. Finally, fill in all the calculation results into the corresponding modules to complete the construction of the equivalent model.

[0028] The equivalent simulation model of the photovoltaic power station constructed by the above method is compared with the detailed simulation model of the photovoltaic power station to verify the accuracy of the equivalent simulation model.

[0029] Figure 2 It is a waveform comparison diagram when the grid connection point export voltage drops to 35% of the rated voltage. Figure 3 It is a waveform comparison diagram of the current at the grid connection point export. Figure 4 It is a waveform comparison diagram of the active power at the grid connection point export. Figure 5 It is a waveform comparison diagram of the reactive power at the grid connection point export. Figures 2 - 5 It shows that the transient responses of the established equivalent model and the detailed model to the external power grid during the low-voltage ride-through period are basically the same, with good consistency. Through calculation, the maximum deviation value of the current in the detailed model and the equivalent model does not exceed 0.0028, the maximum deviation value of the active power does not exceed 0.0012, and the maximum deviation value of the reactive power does not exceed 0.0025, which verifies the high precision of the equivalent model and the effectiveness of the equivalent method.

[0030] The above embodiments are only the preferred technical solutions of the present invention, and should not be regarded as limitations to the present invention. The embodiments in this application and the features in the embodiments can be arbitrarily combined with each other without conflict. The protection scope of the present invention should be the technical solutions recorded in the claims, including the equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, the equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A step-by-step clustering and equivalent modeling method for photovoltaic power stations based on an improved k-means algorithm, characterized in that: The following steps are involved: Step 1: Build a detailed simulation model of the photovoltaic power station, construct the electrical topology of the photovoltaic power station, and enter the electrical parameters of each module; Step 2: Perform power flow calculation and transient calculation on the detailed simulation model of the photovoltaic power station, and collect the power and voltage values ​​during steady state and transient period; Step 3: Use the improved k-means algorithm for clustering: use the active power of the photovoltaic power station during the steady state as an indicator for preliminary clustering, and then use the active and reactive power of the photovoltaic power station during the transient state as indicators for secondary clustering; Step 4: Aggregate the electrical parameters in the detailed simulation model of the photovoltaic power station according to the clustering results, and build an equivalent simulation model of the photovoltaic power station.

2. According to claim 1, a photovoltaic power station step-by-step clustering and equivalent modeling method based on an improved k-means algorithm is characterized in that: In the step 1, the active and reactive power control strategies of the photovoltaic generator set during the low-power run-through period of the detailed simulation model of the photovoltaic power station are as follows: During the steady state, active power is output and reactive power is 0; during the transient state, the active and reactive power control strategies adopted are as follows: The active power control strategy is as follows: （1）； In the formula, K 1_Ip is the active current coefficient 1; K 2_Ip is the active current coefficient 2; Ip set is the active current coefficient 3; Ip LVRT is the active current during the fault period; Ip 0 is the active current before the fault; Vt is the terminal voltage amplitude; The reactive power control strategy is as follows: （2）； In the formula, K 1_Iq is the reactive current coefficient 1; K 2_Iq is the reactive current coefficient 2; Iq set is the reactive current coefficient 3; Iq LVRT is the reactive current during the fault period; Iq 0 is the reactive current before the fault; V L in To enter the low voltage ride-through threshold; Vt is the terminal voltage amplitude.

3. The photovoltaic power station step-by-step clustering and equivalent modeling method based on the improved k-means algorithm according to claim 1 is characterized in that: In the step 2, the Newton-Raphson method is used to calculate the power flow of the photovoltaic power station simulation model.

4. The photovoltaic power station step-by-step clustering and equivalent modeling method based on the improved k-means algorithm according to claim 1 is characterized in that: In the step 2, transient calculation is performed on the photovoltaic power station simulation model: a three-phase fault is set at the grid connection point so that the grid connection point voltage enters a low-breakdown state, and the active and reactive power at the photovoltaic generator set end is collected.

5. The photovoltaic power station step-by-step clustering and equivalent modeling method based on the improved k-means algorithm according to claim 1 is characterized in that: In step 3, the improved k-means algorithm is used to group the photovoltaic power stations. The specific operations are as follows: S301, normalize the grouping indexes in the photovoltaic power station. The primary grouping index is the active power during the steady state, and the secondary grouping index is the active and reactive power during the transient state. Set the number of grouping clusters in the photovoltaic power station to k , and initialize randomly k Cluster centers: （3）； In the formula, For the k A set of initial values ​​for cluster centers; Indicates k The cluster center is i Initial values ​​on feature dimensions; S302, cluster the samples and assign each sample point x i Assigned to the nearest cluster center, x i The weighted Euclidean distance to each cluster center is calculated as: （4）； in, x ij For sample x i No. j eigenvalues; It is a cluster k The center point j eigenvalues; It is a feature j The weight indicates the importance of this feature to the distance calculation; S303, update the cluster center, and calculate the new center of each cluster according to the allocation result: （5）； In the formula, It is k The new center of the cluster is j The value of a feature; C k It is k A sample set of clusters; Compare the change of cluster centers before and after the update. If the change is less than the preset threshold, the cluster center is considered to have converged. S304, calculate the silhouette coefficient, for a sample x i , the silhouette coefficient is defined as: （6）； in, a ( i ) is a sample x i The average distance to other samples in the cluster to which it belongs is expressed as: （7）； In the formula, | μ k | is a cluster μ k The number of samples in ; It is a sample x i and x j The Euclidean distance between b ( i ) is a sample x i The average distance to the nearest other cluster samples is expressed as: （8）； In the formula, It is a cluster The number of samples in ; l ≠ k Representation Cluster l Except x i Other clusters outside the cluster to which it belongs; S305, calculate the overall silhouette coefficient. The overall silhouette coefficient of the clustering result is the average of all sample silhouette coefficients: （9）； During each iteration, record the randomly selected k and w And the corresponding overall silhouette coefficient S ( k , w ), after multiple iterations, S ( k , w ) is the maximum value. k and w .

6. The photovoltaic power station step-by-step clustering and equivalent modeling method based on the improved k-means algorithm according to claim 1 is characterized in that: In step 4, the electrical parameters in the photovoltaic power station are aggregated according to the grouping results. The specific aggregation formula is as follows: S401, the average unit parameters of the photovoltaic power station are as follows: （10）； In the formula, S eq , P eq and Q eq They are the equivalent unit capacity, active power and reactive power in the photovoltaic power station group respectively; S i , P i and Q i Respectively, i The capacity, active power and reactive power of each unit; S402, the equivalent box transformer parameters in the photovoltaic power station are as follows: （11）； Where: S Teq is the equivalent box-type transformer capacity; S Ti For the i The capacity of the platform box changes; Z Teq is the equivalent box transformer impedance; δ i For the i The weight of the capacity of a box transformer in the total box transformer capacity; Z Ti For the i The platform box changes impedance; S403. In a photovoltaic power station, there are two types of collector line topology structures, namely, a trunk structure and a radial structure; When connected in trunk line mode, the equivalent impedance of the collector line is: （12）； In the formula, Z eq is the equivalent impedance of the collector line, Z 1~ Z N The impedance of the busbar collector from the 1st to the Nth generator set to the feeder grid connection point; When connected radially, the equivalent impedance of the collector line is: （13）； In the formula, Z i For the i The impedance of the busbar collector from the generator set to the feeder grid connection point; S404. According to the clustering results, the photovoltaic units in the same group are regarded as an equivalent unit, and the corresponding equivalent transformer and collection line models are built: first, the capacity, active power and reactive power of the units in the cluster are extracted, and the formula of S401 is used to aggregate them into equivalent units; then, the transformer parameters are aggregated, and the aggregation method of S402 is used to ensure that the equivalent transformer capacity, voltage level and impedance are consistent with the corresponding detailed model; then, the collection line is equivalently simplified, and the equivalent impedance of the trunk and radial structures is calculated according to the aggregation formula of S403; finally, all the calculation results are filled into the corresponding module to complete the construction of the equivalent model.

Images