A method for constructing an incident chain search model of power grid cascading failure

Abstract

The application belongs to the technical field of power system security and defense, and provides a method for constructing an accident chain search model of power grid cascading failure, including the following steps: extracting a power supply path associated subnetwork based on a power grid topology structure, the power supply path associated subnetwork being composed of a power supply path where an initial failure element is located and an electrically strongly coupled branch thereof; adopting a link condition function for dynamic pruning; quantifying a failure propagation tendency in combination with an index of a non-homogeneous characteristic of a power grid; correcting a failure propagation probability based on a new energy output fluctuation model; and screening a high-risk accident chain with system splitting as a termination condition, the system splitting being determined according to the power grid being split into two or more than two islands and at least one island losing power balance, the method provided by the application significantly improves the calculation efficiency and risk assessment accuracy.

Description

Technical Field

[0001] This invention belongs to the field of power system security and defense technology, and relates to a method for constructing an accident chain search model for power grid cascading failures. Background Technology

[0002] The accident chain search model for cascading failures in power grids has significant application value in power grid planning, helping planners improve the reliability, security, and stability of the power grid and reduce the occurrence of large-scale power outages. Specific applications of the accident chain search model in power grid planning mainly include the following aspects: 1. Vulnerability Assessment: The accident chain search model can identify weak links in the power grid and assess the system's vulnerability under specific conditions. This helps power grid planners to specifically strengthen these weak links during power grid design and renovation, improving the overall reliability of the power grid. 2. Fault Prediction and Prevention: The accident chain model can help planners predict possible cascading failure paths and formulate corresponding preventative measures by analyzing key equipment and protection devices along these paths. For example, setting reasonable protection settings and operating times can reduce the probability of cascading failures. 3. Power Grid Design and Optimization: During the power grid design phase, the accident chain search model can be used to evaluate the impact of different network topologies on the propagation of cascading failures, helping planners select a better network structure. For example, using small-world network models or scale-free network models can enhance the power grid's resilience to cascading failures. 4. Operation and Maintenance: The incident chain model can be used to monitor the operating status of the power grid online and detect potential cascading failure risks in real time. By monitoring the operating status of critical lines and equipment in real time, planners can take timely preventive measures to avoid cascading failures. 5. Emergency Response Plan Development: Through the incident chain search model, planners can develop detailed emergency response plans, clarifying response measures and recovery procedures under different cascading failure scenarios. This helps to respond quickly in the event of a cascading failure, reducing the impact of the failure on the power grid and society. 6. Risk Assessment and Management: The incident chain model can be used to assess the risk of cascading failures in the power grid. Combined with risk assessment methods, components that may experience cascading failures can be prioritized, thereby optimizing resource allocation and improving the safety and stability of the power grid. 7. Addressing Emerging Power Grid Issues: With the large-scale integration of new energy sources, the complexity and uncertainty of the power grid increase. The incident chain model can be used to analyze emerging power grid issues, such as the impact of the volatility of renewable energy on cascading failures, helping planners develop corresponding response strategies.

[0003] Traditional fault chain search methods employ a full network traversal strategy, resulting in exponentially increasing computational complexity (the combinatorial explosion problem), which is insufficient for online risk assessment. Methods for constructing fault chain search models for power grid cascading faults mainly include: fault chain intermediate link prediction indicators based on grey relational analysis; iterative improvement search patterns; fault chain intermediate link prediction indicators based on network power flow; cascading fault chain identification methods based on quantitative analysis; and cascading fault chain models based on PSASP. While each of these methods has its advantages and disadvantages, their accuracy and reliability are still insufficient. Summary of the Invention

[0004] This invention proposes a method for constructing an accident chain search model for power grid cascading faults, which improves computational efficiency and enhances the accuracy of risk assessment.

[0005] The technical solution of this invention is implemented as follows:

[0006] A method for constructing a fault chain search model for power grid cascading failures includes the following steps:

[0007] Based on the power grid topology, a power supply path association subnet is extracted. The power supply path association subnet consists of the power supply path where the initial fault element is located and its electrically strongly coupled branches.

[0008] Dynamic pruning is adopted using a link condition function, which generates propagation path filtering rules based on component failure rate, latent fault probability of protection device and power flow over-limit threshold, wherein the weight coefficient is dynamically adjusted according to the power grid scale.

[0009] The fault propagation tendency is quantified by combining the heterogeneous characteristics of the power grid, including power flow distribution entropy and topological entropy.

[0010] The fault propagation probability is corrected based on the new energy output fluctuation model, which includes the wind power prediction error distribution and the photovoltaic output stochastic model.

[0011] High-risk accident chains are screened using system disconnection as the termination condition. The system disconnection determination criteria are that the power grid splits into two or more islands and at least one island loses power balance.

[0012] Preferably, the method for constructing the power supply path associated subnet includes:

[0013] The upstream and downstream electrical connection branches of the initial faulty component are traversed by depth-first search;

[0014] Nodes and branches whose power interaction strength with the initial faulty component exceeds a threshold are selected;

[0015] Construct a local network topology subgraph that includes the nodes and branches.

[0016] Preferably, the stage condition function is defined as: P prop =P fault ×P protect ×δ(△P flow ), where P fault P represents the inherent failure rate of the component. protect The probability of the protection device malfunctioning / failing to operate is δ(ΔP). flow () is the trend limit correction factor.

[0017] Preferably, the formula for calculating the power flow distribution entropy is as follows: Among them, P i For the active power flow of the i-th branch, P total This represents the total active power of the entire network.

[0018] Preferably, the method for calculating the topological entropy includes:

[0019] Calculate the connectivity k of each node i ;

[0020] Calculate the dispersion of connectivity distribution: Where p(k) is the percentage of nodes with a connectivity of k.

[0021] Preferably, the new energy output fluctuation model specifically includes:

[0022] The wind farm output prediction error follows a normal distribution with a mean of 0 and a standard deviation of 15% of the predicted value;

[0023] The correlation model between photovoltaic power plant output and solar irradiance is: P PV =η·S·I(t);

[0024] Where η is the conversion efficiency, S is the area of ​​the photovoltaic panel, and I(t) is the real-time irradiance.

[0025] Preferably, the system disconnection determination criteria further include:

[0026] The deviation between the generated power and the load power within the isolated island exceeds ±20%;

[0027] Within the island, the number of nodes with a voltage below 0.7 pu accounts for more than 30%.

[0028] Preferred options also include:

[0029] The risk level of a high-risk accident chain is classified based on the fuzzy comprehensive evaluation method. The evaluation indicators include the fault propagation speed, the amount of impact load, and the recovery time cost.

[0030] Generate a defense strategy library, which includes commands for power cut-off, load shedding, and emergency DC power control.

[0031] Preferably, the membership function of the fuzzy comprehensive evaluation method adopts a trapezoidal distribution, and the risk levels are divided into Level I (urgent), Level II (serious), and Level III (general).

[0032] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described above.

[0033] The working principle and beneficial effects of this invention are as follows:

[0034] 1. The method provided by this invention improves computational efficiency, reduces the search range to 20%-40% of the original power grid scale, and reduces computation time by more than 60%; improves the accuracy of risk identification, and after coupling modeling of non-homogeneous characteristics and new energy fluctuations, the false alarm rate of high-risk accident chains is <5%; the defense strategy is practical, supports the generation of graded early warning and differentiated control instructions, and reduces the probability of large power outages by 30%-50%.

[0035] 2. This invention dynamically limits the search range by associating subnets with power supply paths, and combines link condition function pruning and non-homogeneity characteristic quantification to significantly improve computational efficiency and risk assessment accuracy. It integrates a renewable energy fluctuation model with fuzzy comprehensive evaluation, supports the generation of tiered defense strategies, and is suitable for the security and control of AC / DC hybrid power grids with a high proportion of renewable energy access. Detailed Implementation

[0036] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0037] A method for constructing a fault chain search model for power grid cascading failures includes the following steps:

[0038] Based on the power grid topology, a power supply path association subnet is extracted. The power supply path association subnet consists of the power supply path where the initial fault element is located and its electrically strongly coupled branches. The construction method of the power supply path association subnet includes: traversing the upstream and downstream electrical connection branches of the initial fault element through depth-first search; filtering nodes and branches whose power interaction strength with the initial fault element exceeds a threshold; and constructing a local network topology subgraph containing the nodes and branches.

[0039] Dynamic pruning is employed using a link condition function, which generates propagation path filtering rules based on component failure rate, the probability of latent faults in protection devices, and power flow exceedance thresholds. The link condition function is defined as: P prop =Pfault ×P protect ×δ(△P flow ), where P fault P represents the inherent failure rate of the component. protect The probability of the protection device malfunctioning / failing to operate is δ(ΔP). flow () is the trend limit correction factor.

[0040] The fault propagation tendency is quantified by combining power grid heterogeneity characteristic indicators, including power flow distribution entropy and topological entropy; wherein, the formula for calculating the power flow distribution entropy is as follows: Among them, P i For the active power flow of the i-th branch, P total The total active power of the entire network; the method for calculating the topology entropy includes: calculating the connectivity k of each node. i ; Calculate the dispersion of connectivity distribution: Where p(k) is the proportion of nodes with a connectivity of k;

[0041] The fault propagation probability is corrected based on a new energy power output fluctuation model, which includes a wind power prediction error distribution and a photovoltaic power output stochastic model. Specifically, the new energy power output fluctuation model includes: the wind farm power output prediction error follows a normal distribution with a mean of 0 and a standard deviation of 15% of the predicted value; the correlation model between photovoltaic power plant output and solar irradiance is: P... PV =η·S·I(t); where η is the conversion efficiency, S is the area of ​​the photovoltaic panel, and I(t) is the real-time irradiance; where the standard deviation of wind power prediction error is 15% is based on the statistical analysis of the actual output and prediction data of a wind farm in a certain region from 2020 to 2022 (the error distribution test was passed by the Kolmogorov-Smirnov normality test, p>0.05).

[0042] High-risk accident chains are screened using system disconnection as the termination condition. The system disconnection judgment criteria are that the power grid splits into two or more islands and at least one island loses power balance. Further criteria for system disconnection include: the deviation between the generating power and load power within the island exceeds ±20%; and the proportion of nodes with voltage below 0.7 pu within the island exceeds 30%. The system disconnection judgment criteria reference the islanding operation requirements in the "Guidelines for the Safety and Stability of Power Systems" (GB / T 26399-2011) and verify the rationality of the thresholds using historical fault statistics (2018-2023) of a certain provincial power grid.

[0043] The fuzzy comprehensive evaluation method is used to classify the risk levels of high-risk accident chains. The evaluation indicators include fault propagation speed, impact load, and recovery time cost. A defense strategy library is generated, which includes generator tripping, load shedding, and emergency DC power control commands. The membership function of the fuzzy comprehensive evaluation method adopts a trapezoidal distribution, and the risk levels are divided into Level I (emergency), Level II (severe), and Level III (moderate). Level I risk corresponds to a steep interval of the trapezoidal distribution function (e.g., membership degree ≥ 0.8), Level II is a flat interval (0.5-0.8), and Level III is a low membership degree interval (≤ 0.5).

[0044] Specifically, the steps include the following:

[0045] Step 1: Data Preprocessing and Construction of Related Subnets

[0046] Input data: power grid topology (CIM / E format), real-time SCADA measurement data, and power output prediction curves of renewable energy power plants;

[0047] Power supply path extraction: An improved Dijkstra algorithm is used to search for critical power supply paths from power nodes to load nodes. The path weight is determined by both electrical distance (impedance) and transmission capacity.

[0048] Related subnet generation: Starting from the initial faulty component (such as a 500kV AC line), traverse the three layers of electrical connection nodes upstream and downstream, and select branches with power interaction intensity ≥10% to form a local subnet.

[0049] Step 2: Dynamic pruning of stage condition functions

[0050] Component failure rate P fault Based on the historical fault statistics database, the transformer fault rate is set at 0.002 times / year·unit, and the transmission line fault rate is set at 0.0015 times / year·km.

[0051] Protection device malfunction rate P protect Based on the reliability report of relay protection devices, the false trip rate is set at 0.3% for distance protection and 0.1% for differential protection.

[0052] Power flow overload threshold: A line load rate ≥ 120% is set as an overload, triggering δ(ΔP) flow =1, otherwise 0.

[0053] Step 3: Quantification of Heterogeneous Characteristics and Correction of Propagation Probability

[0054] Example of calculating power flow distribution entropy: A regional power grid contains 120 branches, with a total active power flow P. total =15,000MW, with power flow distribution of a single branch as 500, 300, ..., 800, and H calculated.flow =2.37;

[0055] Application of topological entropy: When H topo When the value is greater than 1.5, the power grid structure is considered to be significantly heterogeneous, and the fault propagation tendency increases by 20%-40%.

[0056] Step 4: Screening of High-Risk Incident Chains and Generation of Defense Strategies

[0057] Termination condition determination: If a certain accident chain leads to two isolated islands, and one of the islands has a power generation deficit of 25%, it is marked as high risk;

[0058] Defense strategy matching: For Level I risk incident chains, prioritize load shedding (ranked by electricity price sensitivity) and emergency DC power increase of 10%.

[0059] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described above.

[0060] Program structure and modules:

[0061] The program includes the following functional modules:

[0062] Data input module: Analyzes power grid topology (CIM / E format), real-time SCADA measurement data, and new energy output prediction curves;

[0063] Related subnet construction module: Based on the depth-first search algorithm, dynamically generate local network topology subgraphs of the initial faulty components;

[0064] Dynamic pruning module: Filters propagation paths based on link condition functions (integrated component failure rate, protection device malfunction probability, and power flow over-limit status);

[0065] Heterogeneous characteristics calculation module: Enables quantitative analysis of power flow distribution entropy and topological structure entropy;

[0066] New energy fluctuation correction module: embedding wind power output normal distribution model and photovoltaic output illumination correlation model;

[0067] Risk assessment module: Performs system disconnection condition detection (islanding power deviation, voltage anomaly, etc.);

[0068] Defense strategy generation module: Outputs hierarchical control commands (shutdown, load shedding, DC power adjustment) based on fuzzy comprehensive evaluation method.

[0069] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0070] It should also be understood that, in the various embodiments of the present invention, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions between different embodiments are consistent and can be referenced by each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0071] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0072] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a fault chain search model for power grid cascading faults, characterized in that, Includes the following steps: Based on the power grid topology, a power supply path association subnet is extracted. The power supply path association subnet consists of the power supply path where the initial fault element is located and its electrically strongly coupled branches. Dynamic pruning is adopted using a link condition function, which generates propagation path filtering rules based on component failure rate, latent fault probability of protection device and power flow over-limit threshold, wherein the weight coefficient is dynamically adjusted according to the power grid scale. The fault propagation tendency is quantified by combining the heterogeneous characteristics of the power grid, including power flow distribution entropy and topological entropy. The fault propagation probability is corrected based on the new energy output fluctuation model, which includes the wind power prediction error distribution and the photovoltaic output stochastic model. High-risk accident chains are screened using system disconnection as the termination condition. The system disconnection determination criteria are that the power grid splits into two or more islands and at least one island loses power balance.

2. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The method for constructing the power supply path associated subnet includes: The upstream and downstream electrical connection branches of the initial faulty component are traversed by depth-first search; Nodes and branches whose power interaction strength with the initial faulty component exceeds a threshold are selected; Construct a local network topology subgraph that includes the nodes and branches.

3. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The conditional function of the process is defined as: P prop =P fault ×P protect ×δ(ΔP flow ), where P fault P represents the inherent failure rate of the component. protect The probability of the protection device malfunctioning / failing to operate is δ(ΔP). flow () is the trend limit correction factor.

4. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The formula for calculating the power flow distribution entropy is as follows: Among them, P i For the active power flow of the i-th branch, P total This represents the total active power of the entire network.

5. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The method for calculating the topological entropy includes: Calculate the connectivity k of each node i ; Calculate the dispersion of connectivity distribution: Where p(k) is the percentage of nodes with a connectivity of k.

6. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The aforementioned new energy output fluctuation model specifically includes: The wind farm output prediction error follows a normal distribution with a mean of 0 and a standard deviation of 15% of the predicted value; The correlation model between photovoltaic power plant output and solar irradiance is: P PV =η·S·I(t); Where η is the conversion efficiency, S is the area of ​​the photovoltaic panel, and I(t) is the real-time irradiance.

7. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The system disconnection determination criteria further include: The deviation between the generated power and the load power within the isolated island exceeds ±20%; Within the island, the number of nodes with a voltage below 0.7 pu accounts for more than 30%.

8. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, Also includes: The risk level of a high-risk accident chain is classified based on the fuzzy comprehensive evaluation method. The evaluation indicators include the fault propagation speed, the amount of impact load, and the recovery time cost. Generate a defense strategy library, which includes commands for power cut-off, load shedding, and emergency DC power control.

9. The method for constructing a fault chain search model for power grid cascading faults according to claim 1, characterized in that, The membership function of the fuzzy comprehensive evaluation method adopts a trapezoidal distribution, and the risk levels are divided into Level I, Level II, and Level III.

10. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 1-9.

Images