A strategy for selecting splitting sections considering splitting fitness indices

Abstract

The present invention discloses a splitting section selection strategy considering splitting fitness index, and the main steps include: S1. Linearize the system including load nodes and generator nodes to determine the regional oscillation mode; S2. Select the modal matrix of the system and construct the correlation matrix; S3. Classify the nodes of the system including load nodes and generator nodes; the nodes are divided into three categories: strongly correlated nodes, weakly correlated nodes and buffer nodes; S4. Construct the splitting applicability index, consider the minimum unbalanced power between regions, and form a comprehensive index; S5. Take the comprehensive index as the objective function for finding the optimal splitting section, and adopt a genetic algorithm to select the optimal splitting section. The splitting section selection strategy of the present invention classifies the system nodes, solves the problem of unclear classification of some original nodes, and reduces the splitting space; by proposing the splitting fitness index, it solves the adaptability of splitting section selection and avoids generator out-of-step after splitting.

Description

Technical Field

[0001] The present invention relates to the technical field of power system stability analysis, and particularly to a sectionalizing section selection strategy considering the sectionalizing applicability index. Background Art

[0002] In recent years, large power outages have occurred frequently in various countries around the world, seriously affecting the normal social life of the local area and causing huge economic losses at the same time. The long-term operation practice of the power system shows that even if the requirements for the stability of the electrical system are getting higher and higher and the safety control scheme is getting more and more perfect, it is still impossible to ensure the transient stability of the power system. When some unpredictable disturbances occur superimposed, if they are not handled reasonably, it will lead to the transient instability of the power system and cause huge economic losses. In order to prevent the out-of-step interconnected power grid from evolving into an uncontrolled large-scale power grid collapse, it is necessary to sectionalize the system into sub-grid networks that maintain synchronous operation at appropriate sections and times.

[0003] With the development of high-speed communication means, active sectionalizing based on online decision-making has been developed. Active sectionalizing can start from the perspective of the system, use high-speed communication means to integrate scattered asynchronous sectionalizing devices, and through real-time, comprehensive and active monitoring of the system state, can timely detect the out-of-step machine groups in the system when a large disturbance occurs, and dynamically determine the sectionalizing points and the action time sequences of each sectionalizing point to prevent the accident from expanding. The online selection of the sectionalizing section in active sectionalizing mainly includes two aspects: section optimization and search. The optimization of the sectionalizing section should comprehensively consider the power angle, voltage and frequency stability of each sub-network after sectionalizing, and try to reduce the control cost after sectionalizing. At present, due to the lack of perfect theoretical support, most existing studies use the minimum unbalanced power (a fast search method for the active sectionalizing section of the power system) and the minimum power flow impact as the objective function to search for the sectionalizing section, so as to improve the stability of the sub-grid network after sectionalizing and minimize the subsequent generator tripping and load shedding. For the scenario where the inertia of the sub-grid network after sectionalizing is small, calculating the optimal sectionalizing section with the above objective function can usually obtain better optimization. However, for large power grids, there may be multiple sectionalizing sections with small differences in exchange power, and the impact on the frequency safety of the sub-grid network after sectionalizing is not much different. It should be noted that in some scenarios, sectionalizing a section with a small sectionalizing exchange power may damage the grid structure of the sub-grid network, form a long-chain power transmission or power supply structure, and bring greater power angle stability risks.

[0004] In the early stage of cross-section search, it was mainly based on graph theory methods. The power system was abstracted as a weighted graph, and the island splitting sets that meet certain constraints were searched (the method for generating power system splitting strategies based on graph theory). As the system scale becomes larger and larger, this method has the problem of the curse of dimensionality and is difficult to meet the requirements of online calculation for large power grids. For example, in the literature: A fast search method for the active splitting cross-section of the power system, based on the multi-layer graph segmentation theory, determines the final splitting cross-section through coarsening, initial partitioning, and reduction optimization, but it is necessary to preset the splitting area in advance and cannot adaptively select the number of final splitting areas.

[0005] With the development of the slow coherency theory, the cross-section search method based on the slow coherency theory has also been developed. For example, in the literature: Research on the expansion planning of the transmission network considering splitting control, an improved method for identifying and screening weak connection lines is proposed based on the slow coherency, which solves the situation that the traditional clustering matrix cannot accurately cluster generators, but the optimization objective is single. In the literature: Adaptive active splitting control based on slow coherency solves the problem of slow coherency clustering of all nodes in the system including load nodes by introducing the generalized eigenvalue analysis technique, but the adaptability of the node classification criterion is insufficient. Summary of the Invention

[0006] Aiming at the problems existing in the existing cross-section search methods, such as unclear classification of some original nodes and the adaptability of cross-section selection, the present invention provides a cross-section selection strategy considering the splitting fitness index.

[0007] The cross-section selection strategy considering the splitting fitness index provided by the present invention is as follows:

[0008] S1. Linearize the system including load nodes and generator nodes to determine the regional oscillation mode. The regional oscillation mode can reflect the dynamics between regions. After the system is disturbed, the inter-regional mode basically remains unchanged.

[0009] S2. Select the modal matrix of the system and construct the correlation matrix. Each column of the correlation matrix represents a splitting area, and S(i,j) represents the degree of correlation between variable i and region group j. The larger the value, the greater the correlation. When the variable is the voltage phase angle, the oscillation correlation between different nodes can be preliminarily judged through the correlation matrix.

[0010] S3. Classify the nodes of the system including load nodes and generator nodes. The system nodes are divided into three categories: strongly correlated nodes, weakly correlated nodes, and buffer nodes:

[0011] The strongly correlated nodes indicate that they are strongly correlated with the nodes in a certain region (for example, region l numbered l) and are hardly correlated with the nodes in other regions. The node is classified into the strongly correlated region l, and the definition formula is as follows:

[0012]

[0013] In the formula, S i* Represents the i-th row element in the correlation matrix S;

[0014] The weakly correlated node means that there is no region that is strongly correlated with it, but there is a relatively correlated region (for example, region k numbered k), and there is only this relatively correlated region, then the node is classified into the relatively correlated region k, and the definition formula is as follows:

[0015]

[0016] In the formula, S ik Represents the element in the i-th row and k-th column of the correlation matrix S;

[0017] The buffer nodes are nodes other than strongly correlated nodes and weakly correlated nodes, which have little correlation with other regions or have similar correlation with two or more other regions. The buffer nodes are not temporarily assigned to any region.

[0018] S4. Construct a split-line suitability index, taking into account the minimum unbalanced power between regions, and form a comprehensive index;

[0019] Among them, the decoupling suitability index reflects the compactness within the region and the distinction between different regions. The decoupling suitability index is as follows:

[0020]

[0021] Where n k is the number of nodes in the kth region; u is the average correlation of all nodes; u k is the mean value of node correlation in the kth region; C k is the node set in the kth region; t is the number of regions, n is the total amount of data in the dataset; x i is the i-th node in the k-th region.

[0022] The comprehensive indicators are as follows:

[0023]

[0024] Where k represents the kth region, t is the total number of divided regions, j represents the jth node under region k, m is the total number of nodes in region k, p represents the load active power, and g is the maximum output active power of the generator.

[0025] S5. Using the comprehensive index as the objective function for finding the optimal sectionalizing plane, the genetic algorithm is adopted to select the optimal sectionalizing plane. In this step, based on the proposed sectionalizing fitness index, the buffer nodes are grouped by the genetic algorithm, with the comprehensive index as the objective function and ensuring that the nodes belonging to the same region are connected to each other as the constraint condition.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] (1) The sectionalizing plane selection strategy considering the sectionalizing fitness index proposed by the present invention classifies the system nodes, solves the problem of unclear classification of some original nodes, and reduces the sectionalizing space.

[0028] (2) By proposing the sectionalizing fitness index, a suitable sectionalizing plane is effectively selected, solving the adaptability problem of sectionalizing plane selection and avoiding the out-of-step of some generator units after sectionalization.

[0029] Other advantages, objectives, and features of the present invention will be partially reflected by the following description and partially understood by those skilled in the art through the research and practice of the present invention. Brief Description of the Drawings

[0030] Figure 1 The flowchart of the sectionalizing plane selection strategy considering the sectionalizing fitness index of the present invention.

[0031] Figure 2 The IEEE-118 bus system diagram in the embodiment.

[0032] Figure 3 The system node classification result diagram in the embodiment.

[0033] Figure 4 The comprehensive index curve diagram.

[0034] Figure 5 The comparison diagram of two sectionalizing planes.

[0035] Figure 6 The comparison diagram of the power angle curves of two sectionalizing schemes.

[0036] Figure 7 The comparison diagram of the power angle curves of Region 1 under two sectionalizing schemes. Detailed Embodiment

[0037] The following is a description of the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention and are not used to limit the present invention.

[0038] As Figure 1 shown, the sectionalizing plane selection strategy considering the sectionalizing fitness index of the present invention includes five steps:

[0039] S1. Linearize the system including load nodes and generator nodes to determine the regional oscillation modes;

[0040] S2. Select the modal matrix of the system and construct the correlation matrix;

[0041] S3. Classify the nodes of the system including load nodes and generator nodes;

[0042] S4. Construct the splitting applicability index, and at the same time consider the minimum unbalanced power between regions to form a comprehensive index;

[0043] S5. Take the comprehensive index as the objective function for finding the optimal splitting section, and use the genetic algorithm to select the optimal splitting section.

[0044] Take Figure 2 the IEEE-118 bus system shown as an example to calculate and verify using the splitting section selection strategy of the present invention.

[0045] The IEEE-118 bus system includes 186 lines and 54 generators. The black boxes in the figure represent generator nodes, and the generators adopt a fifth-order model; the black circles represent load nodes, and the load nodes adopt a constant power model. Build a system linearization model in MATLAB to determine the electromechanical oscillation modes of the system; divide the number of system regions into 3, and obtain the reference variables corresponding to the three regions as the phase angles of node 1, node 80, and node 111 respectively. Define the group where node 1 is located as region 1, and all the generating units in the group form the region 1 units. Define the group where node 80 is located as region 2, and all the generating units in the group form the region 2 units. Define the group where node 111 is located as region 3, and all the generating units in the group form the region 3 units, and then obtain the correlation matrix of the system. The node correlation matrix is shown in Table 1.

[0046] Table 1. Correlation matrix of IEEE-118 bus system

[0047]

[0048]

[0049] Next, according to the node classification criterion proposed in step S3 of the present invention, divide the nodes of the system into strongly correlated nodes, weakly correlated nodes, and buffer nodes. Figure 3 is the obtained system node classification and division result, where the light gray area is the traditional splitting space, and the dark gray is the splitting space obtained by the node classification method proposed by the present invention. It can be clearly seen from the figure that the node classification method of the present invention can effectively reduce the splitting space.

[0050] The light gray area and the dark gray area are respectively used as the splitting spaces to search for sections. Taking the comprehensive index as the objective function, the decline curve of the comprehensive index during the genetic algorithm evolution process is as follows Figure 4 shown. In the figure, (a) is the section search of the traditional splitting space. (b) is the section search of the splitting space of the present invention. It can be seen from the figure that the comprehensive index under the two splitting spaces continuously decreases with the increase of the number of iterations, and finally converges, and the optimal solutions obtained under the two splitting spaces are the same. However, under the traditional splitting space, the search time for the optimal section is 9 s, and the search time under the space given by the present invention is 4 s. The splitting section obtained by the splitting space recommended by the present invention is consistent with that of the traditional splitting space. It shows that the grouping of weakly correlated nodes is accurately determined according to the method proposed by the present invention, and the method proposed by the present invention reduces the number of candidate splitting lines from 67 to 34, and the reduction ratio reaches 49%. This shows that the classification method proposed by the present invention can effectively solve the problem of unclear classification of some nodes and can reasonably reduce the splitting space. The final optimal splitting section is determined, and the splitting lines are: 37-39, 37-40, 34-43, 38-65, 24-70, 72-71, 82-83, 95-96, 96-94, 98-100, 99-100. The splitting section determined by traditional experience is: 38-65, 43-44, 42-49, 70-69, 70-74, 70-75, 84-85, 83-85, 93-92, 94-92, 94-100, 98-100, 99-100. Figure 5 is a comparison chart of the two splitting sections. In the figure, black represents strongly correlated nodes, gray represents weakly correlated nodes, and white represents buffer nodes. The index values of the two schemes are shown in Table 2.

[0051] Table 2. Comparison of indexes of two splitting schemes

[0052]

[0053] To verify the correctness of the selection of the improved splitting section, verification is carried out in the PSASP power system simulation software. The fault is set as: a three-phase grounding short-circuit fault occurs on line 80-81 at 3 s, the fault is removed at 3.5 s, and the system is split at 5 s.

[0054] When the system encounters a large disturbance, the power angle of the system becomes unstable. At 5 s, the network is split according to the traditional splitting scheme, as shown in Figure 6 (a). The units in area 1 cannot maintain the in-area power angle synchronization. At this time, if you want to ensure the stability of area 1, the out-of-step units must be removed, and the splitting effect is not ideal. Figure 6(b) shows the splitting effect of the splitting scheme proposed by the present invention. It can be clearly seen that when the network is split at 5 s, the power angles of the three regions after partitioning are relatively concentrated, and there are no out-of-step units, indicating good clustering effect.

[0055] Next, a simulation is carried out on Region 1 where there are out-of-step units, as Figure 7 shown. (a) represents the power angle curve of Region 1 under the traditional splitting scheme. (b) represents the power angle curve of Region 1 under the optimized splitting scheme. It can be seen that when the network is split after being disturbed, in the traditional scheme, the power angle of Unit 42 suddenly surges around 2 s after splitting, resulting in a serious out-of-step. Unit 40 follows closely and also experiences a serious out-of-step around 3 s after splitting. The power angles of the remaining generators can swing synchronously relatively well. Comparing with the splitting scheme proposed in this paper, all units in Region 1 can remain synchronous after splitting, and no unit experiences an out-of-step phenomenon. In summary, comparing the two splitting schemes, the splitting scheme proposed by the present invention is significantly better than the traditional empirical splitting scheme.

[0056] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the equivalent embodiments with equivalent changes within the scope of the technical solution of the present invention without departing from the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A strategy for selecting a splitting section considering the splitting fitness index, characterized in that, The steps are as follows: S1. Linearize the system including load nodes and generator nodes to determine the regional oscillation modes; S2. Select the modal matrix of the system and construct the correlation matrix; S3. Classify the nodes of the system including load nodes and generator nodes; the nodes are divided into three categories: strongly correlated nodes, weakly correlated nodes, and buffer nodes: The strongly correlated nodes indicate strong correlation with the nodes within a certain area l and have no strong correlation with the nodes in other areas. Such nodes are classified into the strongly correlated l area, and the definition formula is as follows: ； In the formula, represents the elements of the i-th row in the relevance matrix S; The weakly correlated nodes indicate that there is no region strongly correlated with them, but there is a relatively correlated region, the k-region. These nodes are classified into the relatively correlated k-region, and the definition formula is as follows: ； In the formula, represents the element in the \(i\)-th row and \(k\)-th column of the relevance matrix \(S\); The buffer nodes are other nodes except strongly correlated nodes and weakly correlated nodes. The buffer nodes are not temporarily classified into any region; S4. Construct the splitting applicability index, and at the same time consider the minimum unbalanced power between regions to form a comprehensive index; The constructed splitting applicability index is as follows: Wherein, n k is the number of nodes in the k-th region; u is the average relevance of all nodes; u k is the average of the node relevances within the k-th region; C k is the set of nodes within the k-th region; t is the number of regions, n is the total amount of data in the data set; x i is the i-th node within the k-th region; The comprehensive index is as follows: In the formula, k represents the k th area, t is the total number of divided areas, j represents the k th node under area j , m is the total number of nodes within area k , p represents the active power of the load, g is the maximum active power output of the generator; S5. Take the comprehensive index as the objective function for finding the optimal splitting section, and use the genetic algorithm to select the optimal splitting section.

2. The strategy for selecting a splitting section considering the splitting fitness index according to claim 1, characterized in that, In step S5, based on the proposed splitting applicability index, the buffer nodes are grouped using the genetic algorithm, with the comprehensive index as the objective function, and ensuring that the nodes belonging to the same region are connected to each other as the constraint condition.

3. The strategy for selecting a splitting section considering the splitting fitness index according to claim 1, characterized in that, In step S2, for the constructed correlation matrix, each column represents a splitting area, indicating the correlation degree between variable i and area group j. The larger the value, the higher the correlation degree. When the variable is voltage phase angle, the oscillation correlation between different nodes is preliminarily judged through the correlation matrix.

4. The strategy for selecting a splitting section considering the splitting fitness index according to claim 1, characterized in that, In step S1, the regional oscillation modes reflect the dynamics between regions. After the system is perturbed, the inter-regional modes remain unchanged.

Images