A method for steady-state security online assessment of power grids with a high proportion of new energy sources

Abstract

This invention discloses an online steady-state security assessment method for power grids with a high proportion of renewable energy, belonging to the field of power systems. It uses historical data on wind power output, photovoltaic power output, and load from renewable energy bases as input to construct a joint probability distribution of wind power, photovoltaic power, and load over time. After sampling, a large number of scenarios are generated. After retaining some extreme cases in the generated scenarios, clustering is performed to reduce the dimensionality of typical scenario sets and their weights. Weather information is input periodically, and after correcting the scenario set weights, parallel power flow calculations and N-1 security checks are performed on the typical scenario sets in conjunction with power grid topology and line parameters. Based on the static security check results and scenario weights, probabilistic indicators for power grid safe operation are obtained. This invention enables online assessment of the steady-state security of renewable energy power generation bases connected to the power grid in areas with a high proportion of renewable energy, providing a reference for the scheduling of renewable energy-intensive systems.

Description

Technical Field

[0001] This invention belongs to the field of steady-state security verification in power systems, and relates to an online steady-state security assessment method for power grids with a high proportion of new energy sources. Background Technology

[0002] As a crucial energy sector, the power grid is undergoing gradual energy structure improvements, with new energy sources such as wind and solar power playing an increasingly larger role. Western my country, rich in wind and solar resources, is a major region for new energy base development, and the proportion of new energy in the power grid will further increase in the future.

[0003] However, wind power, photovoltaic power and other new energy power generation are mainly affected by weather, and have great randomness and volatility, which adds many uncertainties to the complex power system and challenges the security of the power system.

[0004] Therefore, how to conduct rapid steady-state security verification for power grids with a high proportion of new energy sources and propose probabilistic security indicators to describe power grid security has become a problem worthy of attention. Summary of the Invention

[0005] The technical problem to be solved by this invention is to propose an online steady-state security assessment method for grids with a high proportion of new energy sources, for the case of new energy power generation bases connecting to the grid with strong uncertainty.

[0006] Therefore, the technical solution adopted by the present invention is as follows:

[0007] A method for steady-state security online assessment of power grids with a high proportion of new energy sources, characterized by the following steps:

[0008] Step 1: Obtain historical data on wind power output, photovoltaic power output, and load of the new energy base, and construct the joint distribution of wind power, photovoltaic power, and load over time;

[0009] Step 2: Sample the joint distribution of wind power, photovoltaic power, and load over time to obtain the original scenario set, and select typical scenario sets from it; the original scenario set includes wind power scenario set, photovoltaic scenario set and load scenario set, the wind power scenario set includes historical wind power output curve, the photovoltaic scenario set includes historical photovoltaic power output curve, and the load scenario set includes historical load change curve.

[0010] Step 3: The wind power output curves and photovoltaic power output curves in the typical scenario set are multiplied by Cartesian product and summed to form the power output scenario set to be measured in the new energy base. The load change curves in the typical scenario set are formed to form the load scenario set to be measured in the new energy base. The power output scenario and load scenario combination is constructed, and the power grid topology of the new energy base is calculated in parallel and checked for N-1 safety.

[0011] Step 4: Based on the safety verification results of step S3, calculate the probability index of the steady-state safe operation of the new energy base connected to the power grid.

[0012] Furthermore, in step 2, the method for selecting a typical scene set from the original scene set includes:

[0013] 2.1) Traverse the original scene set. If the output value or load value of a certain scene at a certain moment is greater than the set threshold, then it is regarded as an extreme scene, and the rest are regarded as non-extreme scenes.

[0014] 2.2) Clustering and reducing non-extreme scenes within the original scene set for each type;

[0015] 2.3) Combine the extreme scenarios selected in step 2.1) and the reduced non-extreme scenarios of the same type obtained in step 2.2) to form the typical scenario set for each type.

[0016] Furthermore, step 2.2) specifically refers to:

[0017] K scenes are randomly selected as center points. The remaining scenes are traversed, and the Euclidean distance between each scene and the center point is calculated. The scene is then merged into the set of the nearest centers. The scenes in this center set are averaged to obtain a new center point, and the weight of the new center is increased. This process is repeated until all scenes are assigned to a center. The scenes corresponding to the K centers are used as the reduced non-scene set.

[0018] Furthermore, in step 3, based on the power grid topology and line parameters of the new energy base, one of the PV nodes in the power grid topology is replaced with the output scenario in the combination of output scenario and load scenario, and one or more PQ nodes are replaced with the load scenario in the same combination of output scenario and load scenario. All combinations are traversed, power flow and N-1 safety check are calculated in parallel, and unsafe lines are output.

[0019] Furthermore, during the safety verification of the power grid topology of the new energy base, the output scenario and load scenario combination is equipped with an initial weight value, which is corrected based on the weather information of the future forecast period.

[0020] Furthermore, the method for adjusting the weight values ​​is as follows:

[0021] Obtain weather information for the future forecast period, including wind speed, radiation, and photovoltaic information, and classify the wind speed, radiation, and photovoltaic information.

[0022] For all combinations of output and load scenarios, update the weight value of the combination scenario whose output level is the same as the weather level for the future forecast period. The update formula is as follows:

[0023] w i(t+1)=w i (t)*scale

[0024] Among them, w i (t) represents the weight of the i-th combined scenario at time t, and scale is a scaling factor greater than 1;

[0025] Then, the weights of all combined scenarios are normalized to ensure that the sum of the weights of all combined scenarios is 1.

[0026] Furthermore, the N-1 safety verification of all output scenario and load scenario combinations is independent of each other. Based on the maximum number of computer threads m, parallel power flow calculation and N-1 safety verification of m output scenario and load scenario combinations are performed simultaneously.

[0027] Furthermore, before performing parallel power flow calculations and N-1 security checks on the power grid topology of the new energy base, a step is included to pre-calculate the insecure lines in the power grid topology of the new energy base, specifically:

[0028] 3.1) Select any initial node in the power grid topology, perform a depth-first search on the power grid topology graph, let dfn[u] represent the order in which node u is found by the depth-first search, and low[u] represent the earliest order that node u can be traced back without going through its parent node. Initialize low[u] = dfn[u].

[0029] 3.2) When node u is found, for each node v connected to node u, if node v has not been visited by depth-first search, then perform depth-first search on node v first. If dfn[u] < low[v], it means that node v cannot reach its parent node through a path other than its parent node. The line uv is an unsafe line after being cut, so add the line to the unsafe line set no_test. After the judgment is completed, update low[u] = min{low[u], dfn[v]}.

[0030] If node v has been visited by depth-first search, then directly update low[u] = min{low[u], dfn[v]};

[0031] 3.3) After the priority search is completed, the set of unsafe lines is returned to obtain all unsafe lines. Before performing N-1 safety checks on all output scenarios and load scenario combinations, if a line is in the set of unsafe lines, power flow calculation is not required, and it is directly listed as an unsafe line.

[0032] Furthermore, in step 4, the formula for calculating the probability index of the steady-state safe operation of the new energy base connected to the power grid is as follows:

[0033]

[0034] In the formula, n is the number of combinations of output scenarios and load scenarios, and w i Let num be the weight of the i-th combined scenario. l num represents the total number of lines. unsafe The number of unsafe lines.

[0035] The beneficial effects of this invention are as follows: This invention can generate low-dimensional and highly representative new energy base grid access scenario generation technology, perform parallel power flow calculation, quickly perform safety verification, and obtain the safety probability index of grid operation. Attached Figure Description

[0036] Figure 1 It is the set of key scenarios in the application example of wind power in this invention, which is the original scenario clustering and then the merging of extreme scenarios;

[0037] Figure 2 This is the power grid topology in an application example of the present invention. The red lines represent lines that have been cut off to form islands and lines that have not passed the N-1 safety check.

[0038] Figure 3 This is a flowchart of the present invention. Detailed Implementation

[0039] The present invention will be further described and illustrated below with reference to the accompanying drawings and specific embodiments.

[0040] In one embodiment of the present invention, a steady-state security online assessment method for power grids with a high proportion of renewable energy is provided, the method comprising the following steps:

[0041] S1: Based on the historical data of wind power output, photovoltaic power output, and load of the new energy base, construct the marginal probability distributions of wind power, photovoltaic power, and load. Use the kernel density estimation method to reconstruct the joint distribution of wind power, photovoltaic power, and load over time from the marginal distributions.

[0042] S2: A large set of scenarios is obtained by sampling the joint distribution of wind power, photovoltaic power and load over time. Then, a part of extreme scenarios are separated from the scenario set, and the original scenario set after separation is clustered and reduced. The reduced scenario set is then merged with the extreme scenarios to obtain the typical scenario set.

[0043] The scenario described is a power curve within a day, including wind power scenario, photovoltaic scenario and load scenario. The wind power scenario set includes historical wind power output curves, the photovoltaic scenario set includes historical photovoltaic output curves, and the load scenario includes historical load change curves. In this embodiment, the historical wind power output curve, historical photovoltaic output curve and historical load change curve are obtained within the same historical time period.

[0044] In one specific embodiment of the present invention, before clustering the large number of generated scene sets, it is considered that some extreme operating scenarios, although their probability of occurrence is small, have a significant impact. The clustering process averages the scenes, and these extreme scenarios may be averaged out and become non-extreme scenarios. Therefore, extreme scenarios are extracted separately beforehand, and K-Means clustering is then performed on the remaining scene set. The process is as follows:

[0045] (1) Traverse the original scene set (already standardized). If the output value or load value of a scene at a certain moment is greater than the set threshold T, remove it from the original scene set and add it to the extreme scene set.

[0046] (2) Perform K-Means clustering on the remaining original scene set after the extreme scenes have been separated.

[0047] Randomly select K scenes as center points. Iterate through the remaining scenes, calculate the Euclidean distance between each scene and the center point, and merge that scene into the set of nearest centers. Then, average the scenes in this center set to obtain a new center point and increase its weight. Repeat this process until all scenes are assigned to a center. The scenes corresponding to the K centers form the reduced scene set, which contains K scenes.

[0048] (3) Incorporate the scenes in the extreme scene set into the reduced scene set.

[0049] The wind power scenario, photovoltaic scenario, and load scenario mentioned above are all processed in steps (1)-(3). The wind power output curves in the reduced wind power scenario set and the photovoltaic output curves in the reduced photovoltaic scenario set are Cartesian products and then added together to form the set of output scenarios to be measured for the new energy base. The load change curves in the reduced load scenario set are used as the set of load scenarios to be measured for the new energy base. Each output scenario in the set of output scenarios to be measured is combined with one or more load scenarios in the set of load scenarios to be measured. The weight of each combined scenario is initialized to 1 / n, where n is the number of combinations of output scenarios and load scenarios.

[0050] Based on the power grid topology and line parameters of the new energy base, one PV node in the power grid topology is replaced with the output scenario in the combined scenario, and one or more PQ nodes are replaced with the load scenario in the combined scenario. The combined scenarios are traversed, and the power flow and N-1 safety check are calculated in parallel.

[0051] In this step, the power flow calculation and N-1 verification between combined scenarios are independent of each other. If the maximum number of computer threads is m, then the safety verification of m combined scenarios can be performed simultaneously.

[0052] In one specific embodiment of the present invention, to fully utilize weather forecast information to aid in steady-state safety verification, weather information (wind speed, radiation, photovoltaic) for the next time period is input every 15 minutes, and the values ​​are graded accordingly. The weights of all combined scenarios whose output levels are the same as the weather level for the next time period are updated, and the weight update formula is as follows:

[0053] w i (t+1)=w i (t)*scale

[0054] Among them, w i (t) represents the weight of the i-th combined scene at time t, and scale is a scaling factor greater than 1.

[0055] Then, the weights of all combined scenarios are normalized to ensure that the sum of the weights of all combined scenarios is 1.

[0056] In one specific embodiment of this invention, to reduce computational load, a pre-calculation of the power grid topology of the new energy base is performed first. The Tarjan algorithm is used to detect cut edges in the power grid topology diagram, that is, lines that are definitely unsafe if cut during the N-1 check. The algorithm flow is as follows:

[0057] (1) Select any initial node in the power grid topology and perform a depth-first search (DFS) on the power grid topology graph. Let dfn[u] represent the order in which node u is found by the depth-first search, and low[u] represent the earliest order that node u can be traced back without going through its parent node. Initialize low[u] = dfn[u].

[0058] (2) When node u is found, for each node v connected to node u, if node v has not been visited by depth-first search, then perform depth-first search on node v first. If dfn[u] < low[v], it means that node v cannot reach its parent node through a path other than its parent node. The line uv is an unsafe line after being cut. Add the line to the unsafe line set no_test. After the judgment is completed, update low[u] = min{low[u], dfn[v]}.

[0059] If node v has been visited by depth-first search, then directly update low[u] = min{low[u], dfn[v]}.

[0060] (3) After the priority search is completed, return the set of unsafe lines no_test to obtain all unsafe lines.

[0061] In this embodiment, the unsafe lines in the unsafe line set are edges in the power grid that would form islands after being cut off. When performing N-1 safety checks on the scenario combination, if this edge is in the set no_test, there is no need to perform power flow calculations, and it is directly listed as an unsafe line, reducing the amount of computation.

[0062] S4: Based on the safety verification results of S3, calculate the probability index of steady-state safe operation of the new energy base connected to the power grid:

[0063]

[0064] In the formula, n is the number of combinations of output scenarios and load scenarios, and w i Let num be the weight of the i-th combined scenario. l num represents the total number of lines. unsafe The number of unsafe lines.

[0065] To verify the effectiveness of the present invention, the above method was implemented in a subsequent application example using wind power, photovoltaic, and load data from a certain region in Austria and the IEEE 118 grid structure of PandaPower. The specific steps will not be repeated here, but the technical effects and implementation details are mainly given.

[0066] This case study demonstrates the method described in this invention using Python, calls libraries such as pandapower for solving the problem, and showcases the implementation results for the case data.

[0067] Operating environment:

[0068] Intel Core i3-10105 CPU 3.70GHz, 16GB RAM, Microsoft Windows 10X64

[0069] Python 3.9

[0070] Implementation results:

[0071] This example is based on wind power, solar power, and load data from a location in Austria, spanning one year and with a frequency of 96 points per day. In this example, the Monte Carlo sampling includes 200 scenarios, with T set to a per-unit value of 0.99 for extreme scenarios. For the K-Means clustering method excluding extreme scenarios, K is set to 4, the upper limit for the number of extreme scenarios is set to 2, and the weight correction scale coefficient is set to 1.1. The power grid structure is derived from PandaPower's 118-node example, with the upper and lower limits of the per-unit node voltage set to 0.9 and 1.1 respectively, and the number of parallel threads set to 4.

[0072] Figure 1The study demonstrates the key scenario set of wind power after the original scenario clustering and then merging of extreme scenarios. It can be seen that even if some scenarios are averaged by the clustering algorithm, some extreme cases with smaller weights are still retained, reflecting extreme operating scenarios with low probability but large impact.

[0073] Clustering results: 5 wind power scenarios, 6 photovoltaic scenarios, and 5 load scenarios were obtained. The Cartesian product of the wind power and photovoltaic scenarios yielded 30 typical output scenarios. The 30 typical output scenarios were then combined with the 5 load scenarios to obtain 30 combined scenarios.

[0074] Figure 2 The power grid topology is shown. The bolded lines are those that have been cut off to form islands and those that have not passed the N-1 safety check. Based on these results, the probability index of safe operation of the power grid is calculated.

[0075] The program output is as follows:

[0076] The final number of scenarios involved in the verification 30 Probability of safe operation 99.31％ Security check time 40.91s

[0077] The results of this case demonstrate that the online steady-state safety assessment method for power grids with a high proportion of new energy proposed in this invention can generate a set of low-dimensional and representative steady-state operation scenarios for power systems in the case of new energy power generation bases connected to the power grid with strong uncertainties. It can also perform a rapid steady-state safety verification of the scenarios within 1 minute and calculate the probabilistic indicators of power grid safe operation.

[0078] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. Rather, the invention encompasses any alternatives or equivalent methods made within the scope of the claims as defined herein. Furthermore, to provide a better understanding of the invention, certain specific details are described below. However, those skilled in the art will fully understand the invention even without these detailed descriptions.

Claims

1. A method for steady-state security online assessment of a power grid with a high proportion of new energy sources, characterized in that, Includes the following steps: Step 1: Obtain historical data on wind power output, photovoltaic power output, and load of the new energy base, and construct the joint distribution of wind power, photovoltaic power, and load over time; Step 2: Sample the joint distribution of wind power, photovoltaic power, and load over time to obtain the original scenario set, and select typical scenario sets from it; the original scenario set includes wind power scenario set, photovoltaic scenario set and load scenario set, the wind power scenario set includes historical wind power output curve, the photovoltaic scenario set includes historical photovoltaic power output curve, and the load scenario set includes historical load change curve. Methods for selecting typical scene sets from the original scene set include: 2.1) Traverse the original scene set. If the output value or load value of a certain scene at a certain moment is greater than the set threshold, then it is regarded as an extreme scene, and the rest are regarded as non-extreme scenes. 2.2) Clustering and reducing non-extreme scenes within the original scene set for each type; 2.3) Combine the extreme scenarios selected in step 2.1) with the reduced set of non-extreme scenarios of the same type obtained in step 2.2) to form the typical scenario set for each type; Step 3: The wind power output curves and photovoltaic power output curves in the typical scenario set are multiplied by Cartesian product and summed to form the power output scenario set to be measured in the new energy base. The load change curves in the typical scenario set are formed to form the load scenario set to be measured in the new energy base. The power output scenario and load scenario combination is constructed, and the power grid topology of the new energy base is calculated in parallel and checked for N-1 safety. Step 4: Based on the safety verification results of step S3, calculate the probability index of the steady-state safe operation of the new energy base connected to the power grid.

2. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 1, characterized in that, Step 2.2) specifically refers to: Randomly select K scenes as center points, traverse the remaining scenes, calculate the Euclidean distance between each scene and the center point, merge the scene into the nearest center set, and average the scenes in this center set to obtain a new center point, and increase the weight of the center; repeat until all scenes are assigned to a certain center, and use the scenes corresponding to the K centers as the reduced non-scene set.

3. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 1, characterized in that, In step 3, based on the power grid topology and line parameters of the new energy base, one of the PV nodes in the power grid topology is replaced with the output scenario in the combination of output scenario and load scenario, and one or more PQ nodes are replaced with the load scenario in the same combination of output scenario and load scenario. All combinations are traversed, power flow and N-1 safety check are calculated in parallel, and unsafe lines are output.

4. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 3, characterized in that, During the safety verification of the power grid topology of the new energy base, the output scenario and load scenario combination are equipped with an initial weight value, which is corrected according to the weather information of the future prediction period.

5. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 4, characterized in that, The method for adjusting the weight values ​​is as follows: Obtain weather information for the future forecast period, including wind speed, radiation, and photovoltaic information, and classify the wind speed, radiation, and photovoltaic information. For all combinations of output and load scenarios, update the weight value of the combination scenario whose output level is the same as the weather level for the future forecast period. The update formula is as follows: ； in, For time t, the first The weights of each combined scenario, A proportionality coefficient greater than 1; Then, the weights of all combined scenarios are normalized to ensure that the sum of the weights of all combined scenarios is 1.

6. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 1, characterized in that, The N-1 safety verification of all output scenario and load scenario combinations is independent of each other. Based on the maximum number of computer threads m, parallel power flow calculation and N-1 safety verification of m output scenario and load scenario combinations are performed simultaneously.

7. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 1, characterized in that, Before performing parallel power flow calculations and N-1 security checks on the power grid topology of the new energy base, the process also includes a step of pre-calculating insecure lines in the power grid topology of the new energy base, specifically: 3.1) Select any initial node in the power grid topology, perform a depth-first search on the power grid topology graph, and use... Represents a node The order in which they are found by depth-first search. Represents a node The earliest order that can be traced without going through the parent node, initialization. ; 3.2) Nodes found At that time, for nodes Each connected node If node If a node has not been visited by depth-first search, then search for it first. Perform a depth-first search, if... This indicates that the node The parent node cannot be accessed through a path other than its parent node. If a line is found to be unsafe after being cut, add it to the unsafe line set no_test; after the judgment is complete, update... ; If node If it has been accessed by depth-first search, update it directly. ; 3.3) After the priority search is completed, the set of unsafe lines is returned to obtain all unsafe lines. Before performing N-1 safety checks on all output scenarios and load scenario combinations, if a line is in the set of unsafe lines, power flow calculation is not required, and it is directly listed as an unsafe line.

8. The steady-state security online assessment method for a power grid with a high proportion of new energy sources according to claim 1, characterized in that, In step 4, the formula for calculating the probability index of the steady-state safe operation of the new energy base connected to the power grid is as follows: ； In the formula, This represents the number of combinations of output and load scenarios. For the first The weights of each combined scenario, The total number of lines, The number of unsafe lines.

Images