Power distribution network multi-scale toughness evaluation method based on multi-disaster coupling modeling

Abstract

The invention discloses a power distribution network multi-scale toughness evaluation method based on multi-disaster coupling modeling, and relates to the field of power system optimization scheduling, in particular to an active power distribution network toughness improvement method based on deep reinforcement learning. The method comprises the following steps: (1) establishing a typhoon wind field model of a region where the power distribution network is located; (2) establishing a typhoon path model of a region where the power distribution network is located; (3) power distribution network fault scene simulation; and (4) carrying out load reduction based on the load importance degree. (5) constructing a toughness evaluation index system; (6) based on the fuzzy entropy, carrying out weight calculation on the toughness indexes of all levels; and (7) carrying out adaptive combination weighting fusion.

Description

Technical Field

[0001] This invention relates to the field of power system optimization scheduling, specifically a multi-scale resilience assessment method for distribution networks based on multi-hazard coupling modeling. Background Technology

[0002] With the frequent occurrence of extreme events such as natural disasters, terrorist attacks, and severe technical failures, power systems face unprecedented challenges in terms of safety and stability. Traditional assessment methods often lack comprehensiveness and dynamism when responding to disaster impacts, making it difficult to accurately reveal the vulnerable links of the distribution network under extreme conditions. To address this issue, resilience assessment strategies are particularly important. This method provides a systematic reference framework for analyzing the impact of natural disasters on distribution network operation, ensuring that the assessment covers key elements in distribution network operation and can be quantitatively characterized. Through scientific indicator selection and design, the accuracy of the results can be improved while maintaining objectivity, thus providing more valuable technical support for decision-making departments. Furthermore, a reasonable evaluation system can not only identify potential weaknesses in the distribution network during disaster prevention and recovery but also provide direction for improving the overall resilience of the system and optimizing emergency management methods. The core idea of ​​this invention is to establish a power distribution network scenario model under typhoon disasters, including wind field and path models; based on this, Monte Carlo sampling is used to estimate the failure rate of system components, thereby simulating different fault scenarios; at the same time, load importance is combined to perform graded reduction, and a multi-scale resilience index system is constructed; finally, a comprehensive evaluation is carried out through a combined weighting method, thereby quantifying the adaptability and recovery capability of the power distribution network under extreme disaster conditions. Summary of the Invention

[0003] Objective: This invention aims to propose a multi-scale resilience assessment method for distribution networks based on multi-hazard coupling modeling. By introducing multi-dimensional assessment indicators, it considers both the changing characteristics of the distribution network's power supply capacity during disasters and the regional differences in the external environment, establishing a comprehensive evaluation model that reflects the overall resilience level of the system. This method utilizes a combined weighting mechanism to effectively balance the correlation and objectivity among indicators, thereby improving the scientific rigor and reliability of the assessment results. Compared with existing technologies, this invention not only more comprehensively depicts the dynamic impact of disaster shocks on the distribution network but also provides quantitative evidence for resilience improvement and emergency decision-making, ultimately enhancing the distribution network's defense and recovery capabilities in the face of extreme disasters.

[0004] Technical solution: The specific steps of this invention are as follows: S1: Establish a typhoon wind field model for the area where the power distribution network is located. Make reasonable assumptions that the typhoon intensity remains constant within this time range. Based on the Batts wind field model, use the superposition of gradient wind speed in the wind circle and typhoon movement speed to simulate the wind speed at each point within the influence range of the typhoon wind circle.

[0005] S2: Establish a typhoon path model for the region where the power distribution network is located. Taking into full account the geographical location of the power distribution network and historical typhoon path information, a combination of probability distribution fitting of historical typhoon path data and Monte Carlo random sampling is adopted to construct a typhoon disaster path model.

[0006] S3: Distribution network fault scenario simulation based on multi-hazard coupling modeling. It integrates the coupling relationship between typhoon, rainfall, terrain and equipment health status, uses model Carlo sampling to randomly simulate distribution network fault scenarios under extreme events, and uses mathematical statistics to solve the selected distribution network fault characteristics.

[0007] S4: Load reduction based on load importance. Based on the importance analysis of distribution network loads and the different power supply reliability requirements of different load levels, the loads at each node are classified into Level 1, Level 2, and Level 3 loads. Load reduction schemes are then determined for each load level.

[0008] S5: Selection of Resilience Assessment Indicators. This considers changes in the internal power supply capacity of the distribution network during a disaster, as well as the regional characteristics of the external disaster. For this type of complex decision-making problem involving multiple criteria, the analytic hierarchy process (AHP) is used to hierarchically classify the assessment indicators, establishing a hierarchical resilience assessment indicator system.

[0009] S6: Calculate Objective Weights Based on Fuzzy Entropy. To comprehensively reflect the objective differences and uncertainties of resilience assessment indicators, this step uses the fuzzy entropy method to quantify the information content of the indicators and calculate objective weights. The fuzzy entropy method can more accurately characterize the information contribution of indicators when the sample data has fuzziness, uncertainty, or nonlinear characteristics.

[0010] S7: Adaptive Combined Weighting and Fusion. Under complex disaster conditions, the importance of each indicator is dynamically adjusted according to time and the stage of the disaster. By combining real-time data during the disaster evolution process, the weights are automatically adjusted according to the stage.

[0011] In step S1, the wind field distribution of the power distribution network area is first constructed using the Batts typhoon wind field model. Since the research object is a regional power distribution network, the duration of its impact by the typhoon is relatively limited; therefore, it can be reasonably assumed that the typhoon intensity remains constant during this period. This wind field model characterizes the wind speed distribution at various locations within the typhoon's wind circle by superimposing the gradient wind speed within the wind circle with the typhoon's moving speed. Friction effects are not considered in this process. The formula for calculating the gradient wind speed is as follows: In the formula: The maximum gradient wind speed (km / h) within the wind circle; K is an empirical coefficient with a value of 6.72; The central pressure difference (hpa); The radius of the typhoon's maximum wind speed (km); The Coriolis force coefficient for Earth's rotation is taken as the geographical latitude of 30°.

[0012] Based on this, through gradient wind speed and typhoon movement speed By superimposing the data, we can obtain the average maximum wind speed within the typhoon's wind circle at a height of 20m above the sea surface within 10 minutes. The calculation formula is as follows: Furthermore, when any point on the sea surface is a distance *r* from the typhoon center, the typhoon wind speed at that location is: In the formula: Let r be the average wind speed (km / h) at any point in the typhoon wind field at a distance r from the typhoon center. This represents the average maximum wind speed within the typhoon's wind circle. 7r represents the radius of the typhoon's maximum wind speed (km); 7r represents the distance from the research point to the typhoon's center (km).

[0013] After a typhoon makes landfall, the reduction effect of terrain roughness (such as mountains, basins, or building complexes) on wind speed needs to be considered. Therefore, a terrain correction coefficient is introduced into the model to adjust for wind speed, resulting in the formula for calculating the 10-minute average wind speed at a height of 10m on land: In the formula: is the average wind speed (km / h) at any point in the typhoon wind field on land. denoted as the average wind speed (km / h) at any point in the typhoon wind field over the sea; p is the obstacle factor, taken as 0.85; z is the ground roughness length, which is taken as 1 according to the roughness classification standard since the study focuses on urban power distribution networks.

[0014] In step S2, a typhoon track model for the study area needs to be established. This model must reflect both the historical typhoon movement characteristics and the geographical conditions of the target power distribution network location. The specific process is as follows: (1) Define the simulation area. Delineate the radius with the study point as the center. The simulated circle is used to determine whether a typhoon will affect the study point. The minimum distance between the study point and the typhoon's path... Smaller than the radius of the simulated circle They believed that the typhoon would affect the research site; conversely, it would not have any impact.

[0015] (2) Historical data screening and preprocessing: To ensure the reasonableness of the probability distribution fitting, only typhoons, strong typhoons, and super typhoons with wind speeds exceeding level 12 at landfall were retained. At the same time, it was ensured that the parameter values ​​were within a reasonable range: (3) Probability distribution fitting of key parameters: Statistical modeling and distribution fitting of key parameters such as the typhoon's moving speed, wind direction, and minimum distance from the research point to the typhoon's path.

[0016] (4) Typhoon path generation: By using Monte Carlo random sampling, a set of parameters (moving speed, wind direction and minimum distance) are extracted from the fitted probability distribution to construct a virtual typhoon path.

[0017] (5) Multi-path expansion: If multiple virtual typhoon paths need to be generated, the sampling and construction process is repeated to form a set of typhoon paths for multiple scenarios.

[0018] The typhoon path model established through the above steps can introduce uncertainty modeling while taking into account historical statistical characteristics, laying the foundation for subsequent disaster scenario simulation.

[0019] In step S3, to realistically depict the comprehensive impact of natural disasters on the structure and equipment of active power distribution networks, this invention first establishes a multi-hazard coupling model considering typhoons, rainstorms, terrain, and equipment health status. This model, by introducing the spatiotemporal superposition and coupling effects of disaster factors, achieves a full-chain representation from environmental impact to equipment damage and then to system functional degradation. Monte Carlo sampling is used to generate power distribution network fault scenarios under disaster conditions, and system resilience indicators are calculated accordingly. The method is as follows: 1. Modeling the Multi-Hazard Complex Mechanism: First, a probabilistic model is constructed to describe the impact of disasters, specifically the multi-dimensional effects of typhoons and rainstorms on active power distribution networks. This model addresses the following: typhoons generate high wind speeds, causing mechanical damage to overhead lines and tower structures; rainstorms increase soil moisture, leading to changes in grounding resistance and equipment insulation degradation; topography and geomorphological features influence the distribution of wind and rain fields and the path of disaster propagation; and equipment aging and health status determine differences in vulnerability. The parameters represent the operating status and failure probability of power distribution network components. A multi-hazard coupling strength function is introduced. The comprehensive representation of the node at time 1 Intensity of the disaster impact: in, For nodes Wind speed, For nodes At the intensity of rainfall, For nodes Terrain exposure coefficient The device health index is [0,1], the smaller the value, the more vulnerable the device. , , , For disaster factor weighting coefficients, satisfying .

[0020] 2. Analysis of the Impact of Electrical Degradation: Heavy rain can lead to water immersion and insulation degradation in equipment. This invention uses a modified log-normal rainfall model to describe the rainfall intensity distribution in different regions: in, This represents the maximum rainfall intensity. The distance from the node to the center of the rainstorm; This is a time-dependent disturbance factor used to reflect the change in rainfall intensity over time. The insulation degradation rate caused by rainfall is expressed as: in The aging rate of equipment under normal conditions. This is the rainfall sensitivity coefficient. 3. Equipment cumulative damage model and node failure rate calculation Based on equipment health This reflects its remaining performance under disaster conditions. Based on historical equipment operation and maintenance data and disaster stress levels, a degradation differential equation is defined: in This represents the vulnerability coefficient related to the device type. Integral yields: when When this occurs, the device is considered faulty.

[0021] Failure rate of nodes or lines Comprehensive Disaster Intensity Logical relationship: in This is a spatial coupling term used to describe the fault propagation effect between adjacent nodes.

[0022] in The disaster propagation coefficient, Let the delay effect of node j failure on node i be the time. 4. Disaster scenarios are generated based on the spatiotemporal coupling characteristics of disasters. By integrating the coupling relationship between typhoons, rainfall, terrain and equipment health status, a multi-disaster scenario set is generated using the Monte Carlo sampling method. Each scenario includes wind speed, rainfall intensity, failure rate and equipment status at different time series.

[0023] The overall disaster intensity distribution across different scenarios can be represented as follows: By generating random numbers using a computer and determining whether components are faulty according to a probability distribution, the operating state of the power distribution network at a given moment can be obtained. If the total number of components is N, its operating state can be represented by a vector: Operating status of each component It can be determined by the uniformly distributed random number r and the fault probability p: Multiple sampling: Repeat the above process M times to obtain M power distribution network system state samples, which represent various fault scenarios that may occur under extreme disasters. Convergence and Accuracy Criteria: According to the law of large numbers, as the number of samplings increases, the simulation results will gradually approach the true values. Combining this with the central limit theorem, the variance coefficient can be used as a convergence criterion. Topology Analysis: When sampling results show that a component has failed and is out of service, the topology of the distribution network needs to be recalculated to determine whether connectivity and power flow distribution have changed. The distribution network consists of nodes and switching lines, and graph theory methods can be used for topology analysis to verify node reachability and system partitioning. When sampling shows a component in the system as faulty, that component is out of service. This may change the connectivity of the original distribution network, thus affecting the power flow distribution. Therefore, topology analysis is needed for the system after the fault to determine the number of nodes and the connection status of each node. The distribution network consists of nodes and switching lines, and graph theory can be used for topology analysis. Solving graph theory problems often requires graph search, which involves starting from a vertex and sequentially visiting the remaining vertices of the graph, with each vertex visited at most once. The basic idea is to select a vertex and mark it, then search for its neighbors and mark them, and continue searching for the neighbors of those neighbors until all vertices in the graph have been marked.

[0024] In step S4, the concept of load importance is introduced to design load reduction strategies. Different nodes have varying degrees of dependence on system power supply reliability; therefore, they need to be classified according to importance, and reasonable load reduction schemes are formulated based on the classification results. Specifically, the importance of each load in the distribution network is first analyzed, and considering the differentiated power supply reliability requirements of different load levels, all node loads are divided into three levels: Level 1, Level 2, and Level 3 loads. Then, based on the classification results, corresponding load reduction strategies are formulated for different load levels.

[0025] Load classification principles and application scenarios: Based on the requirements for power supply continuity, node loads are divided into three levels: Primary loads require a continuous power supply under all circumstances. Typical application environments include: 1) Places where a power outage would directly cause a life-threatening accident, such as hospitals at level two or above; 2) Industries where power outages would cause serious political or economic losses, such as product processing based on important raw materials, steel smelting, rocket launch sites, and fire protection power supply for high-rise buildings; 3) Key power consumption locations that have a significant impact on the political and economic situation, such as important transportation hubs, star-rated hotels, large conference and exhibition centers, and important public facilities.

[0026] For primary loads, power supply solutions include dual power supply and dedicated power lines. Meanwhile, distributed generation provides another convenient means to ensure continuous power supply to these loads. When a distribution network fails, distributed generation can quickly supply power to primary loads within its capacity. Although this solution is slightly less economical, it can significantly reduce investment in distribution network equipment, and its overall benefits remain good after optimization.

[0027] Secondary loads are required to maintain power supply as much as possible, but partial interruptions are permissible in extreme cases. Common scenarios include: 1) Enterprises that would have a significant political or economic impact if a power outage occurs, such as continuous production enterprises; 2) Power outages can disrupt the normal operation of important public facilities, such as transportation hubs, large shopping malls, large stadiums and theaters.

[0028] For these types of loads, dual-circuit power supply and dual-transformer power supply are adopted. Meanwhile, distributed generation can also serve as a supplementary measure. In the event of a fault in the distribution network, distributed generation can provide timely power support to some or even all of the secondary loads, provided that capacity allows.

[0029] Level 3 loads do not have strict requirements for power supply continuity. Typical applications include rural residential electricity consumption and most township enterprise electricity consumption, as well as ordinary loads that do not belong to Level 1 or Level 2 loads. For these types of loads, distributed power sources can be used as random power sources to fully realize their economic, social, and environmental benefits. It should be emphasized that load classification is relative and should be reasonably divided in conjunction with the local power security level and taking into account political and economic impacts.

[0030] Since the resilience of a distribution network is often reflected in its ability to support and recover critical loads, different weights should be assigned to different load levels during the assessment: Level 1 loads are assigned a value of 6, Level 2 loads 3, and Level 3 loads 1. This differentiated weighting can more intuitively reflect the influence of various load types in the resilience assessment.

[0031] When a system failure occurs and topology analysis reveals line overload, load shedding must be performed to restore the safe and stable operation of the distribution network. During this process, the goal of load shedding should be to minimize the amount of load removed while ensuring system safety, and to prioritize preventing the shedding of critical loads. The specific steps are as follows: 1) In the distribution network topology after the fault, each branch is processed in layers starting from the power source node, and then searched one by one in the order of the layers. 2) When an overload is detected on a line, first determine the magnitude of the overload power; 3) Select appropriate load shedding combinations based on load levels to ensure that the total amount of load shedding is not less than the overload power. When implementing load shedding, follow the principle of "from low to high": first shedding level 3 loads; if this is still insufficient to eliminate the overload, then shedding level 2 loads in sequence, and finally considering level 1 loads. 4) Continue searching the system according to the topology hierarchy. If other lines are found to be overloaded, repeat steps 2) and 3) until all lines in the system are restored to a safe operating state.

[0032] In step S5, a resilience assessment index system is constructed. This system must reflect both the dynamic changes in the internal power supply capacity of the distribution network after a disaster and the regional characteristics of the external disaster. To address this type of complex decision-making problem involving multiple criteria, this invention employs the Analytic Hierarchy Process (AHP) to establish a multi-scale resilience index framework by hierarchically classifying the indicators. The entire index system is divided into three levels: (1) Target layer: that is, to assess the overall resilience level of the distribution network.

[0033] (2) First-level indicator layer: It forms the framework of the resilience assessment system and covers core indicators in different dimensions.

[0034] (3) Secondary indicator layer: refined to specific measurement parameters.

[0035] Table 1 Resilience Assessment Index Framework The specific contents of this indicator system are as follows: (1) Distribution network power supply capacity: including the following three aspects: missing area of ​​system function curve, disaster absorption rate and post-disaster recovery rate.

[0036] The missing area of ​​the system function curve can be measured by plotting dynamic function curves before, during, and after the disaster. The load importance index is defined as the load importance weight at the current time t. With loss of load The product of is calculated using the following formula: Therefore, the missing area of ​​the real-time function curve and the system function curve considering load importance can be expressed as: The disaster absorption rate index of a distribution network represents the ratio of the load level that the system can maintain during a disaster to the initial load level before the disaster. It is used to characterize the distribution network's ability to withstand external shocks. The calculation formula is as follows: In the formula: The load level that can maintain continuous power supply during a disaster; This represents the initial load level of the system.

[0037] The disaster recovery rate of a distribution network measures how quickly a distribution network can restore its initial power supply level after a disaster. The formula for its calculation is: In the formula: This represents the missing values ​​for the load level of the distribution network during the disaster. The time it takes for a power distribution network to recover from a load loss to the restoration of full load power supply.

[0038] (2) The degree of impact of disasters on the power distribution network is mainly described by the following three aspects: the frequency of typhoon disasters, the average wind speed of typhoons, and the duration of typhoon impact on the power distribution network. The formula for calculating the duration of typhoon impact T is as follows: In the formula: The radius of the simulated circle in the typhoon model; This is the minimum distance from the geographical center of the power distribution network to the typhoon's path.

[0039] In step S6, to comprehensively reflect the objective differences and uncertainties of resilience assessment indicators, this step uses the fuzzy entropy method to quantify the information content of the indicators and calculate objective weights. Compared with the traditional entropy weight method, the fuzzy entropy method can more accurately characterize the information contribution of the indicators when the sample data has fuzziness, uncertainty, or nonlinear characteristics. The specific method is as follows: 1) Constructing a data matrix It has Each evaluation indicator and For each evaluated object, the raw data for each indicator can be represented by a matrix, where each row of the matrix represents the number of evaluated objects. The column represents the number of evaluation indicators. : 2) Data standardization processing For data matrix Standardization is performed to obtain a standardized data matrix. The processing method for each indicator can be selected according to the indicator type. The specific methods are as follows: Profitability metrics: Cost indicators: 3) Membership function calculation: This reflects the uncertainty of the sample by mapping each standardized index value to the "degree of membership to the ideal state." Membership is defined as: Meanwhile, considering the impact of indicator ambiguity, adjustment parameters are introduced. , represents the fuzzy recognition sensitivity of the indicator, and the improved membership degree is: in Let be the mean membership degree of the j-th index. This treatment helps mitigate the distortion effect of extreme samples on the overall weight distribution.

[0040] 4) Fuzzy entropy calculation: Fuzzy entropy value The formula for calculating the j-th index reflects the uncertainty of information. When the indicator samples have large differences and high information content It has a wide distribution in the [0,1] interval, leading to Smaller, when sample differences are small and information content is low. Concentrated distribution Larger.

[0041] 5) Calculation of objective weights The fuzzy entropy weights are obtained by normalizing the effective information content: This weight reflects the objective contribution of each indicator to changes in system resilience, effectively avoiding the oversensitivity of the traditional entropy weight method to extreme samples. The calculated... This will serve as an objective benchmark for subsequent adaptive combination weighting and fusion.

[0042] In step S7, under complex disaster conditions, the importance of each indicator dynamically adjusts with time and disaster stage. To address this, an adaptive combined weighting algorithm based on fuzzy entropy-Bayes fusion is proposed. This algorithm combines real-time data from the disaster evolution process to achieve self-adjustment of weights with different stages, and constructs a multi-scale resilience dynamic assessment framework. The specific process is as follows: 1) Dynamic update of Bayesian subjective weights: in: Let t be the explanatory probability of the j-th index for changes in system resilience. This represents the subjective weighting of the previous stage.

[0043] This formula allows the indicator weights to be adaptively updated based on real-time disaster observation data. For example, if changes in the system function curve recovery rate during the disaster phase significantly affect the resilience results, then the weights of this indicator... Increase Automatic ascent reflects a dynamic subjective correction mechanism.

[0044] 2) Adaptive weight fusion: To comprehensively consider both the fuzzy entropy weights of objective data and the subjective Bayesian weights, a time-varying adjustment factor is defined. : in: Let be the standard deviation of the fluctuation of indicator j within the time window; This represents the average standard deviation of the volatility of all indicators; when the indicators are highly volatile (information changes rapidly), When the value increases, the model tends to use objective entropy weights; when the indicator changes slowly or is dominated by experience, the model tends to use objective entropy weights. The reduction relies more on subjective Bayesian weights.

[0045] 3) Calculation of combined weights: The final adaptive combined weights are defined as follows: Attached Figure Description

[0046] Figure 1 Flowchart of Multi-Scale Resilience Assessment for Distribution Networks Figure 2 Disaster Failure Scenario Simulation Flowchart Figure 3 Resilience assessment flowchart Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0048] A multi-scale resilience assessment method for distribution networks based on multi-hazard coupling modeling includes the following steps: S1: Establish a typhoon wind field model for the area where the power distribution network is located. Make reasonable assumptions that the typhoon intensity remains constant within this time range. Based on the Batts wind field model, use the superposition of gradient wind speed in the wind circle and typhoon movement speed to simulate the wind speed at each point within the influence range of the typhoon wind circle.

[0049] S2: Establish a typhoon path model for the region where the power distribution network is located. Taking into full account the geographical location of the power distribution network and historical typhoon path information, a combination of probability distribution fitting of historical typhoon path data and Monte Carlo random sampling is adopted to construct a typhoon disaster path model.

[0050] S3: Distribution network fault scenario simulation based on multi-hazard coupling modeling. It integrates the coupling relationship between typhoon, rainfall, terrain and equipment health status, uses model Carlo sampling to randomly simulate distribution network fault scenarios under extreme events, and uses mathematical statistics to solve the selected distribution network fault characteristics.

[0051] S4: Load reduction based on load importance. Based on the importance analysis of distribution network loads and the different power supply reliability requirements of different load levels, the loads at each node are classified into Level 1, Level 2, and Level 3 loads. Load reduction schemes are then determined for each load level.

[0052] S5: Selection of Resilience Assessment Indicators. This considers changes in the internal power supply capacity of the distribution network during a disaster, as well as the regional characteristics of the external disaster. For this type of complex decision-making problem involving multiple criteria, the analytic hierarchy process (AHP) is used to hierarchically classify the assessment indicators, establishing a hierarchical resilience assessment indicator system.

[0053] S6: Calculate Objective Weights Based on Fuzzy Entropy. To comprehensively reflect the objective differences and uncertainties of resilience assessment indicators, this step uses the fuzzy entropy method to quantify the information content of the indicators and calculate objective weights. The fuzzy entropy method can more accurately characterize the information contribution of indicators when the sample data has fuzziness, uncertainty, or nonlinear characteristics.

[0054] S7: Adaptive Combined Weighting and Fusion. Under complex disaster conditions, the importance of each indicator is dynamically adjusted according to time and the stage of the disaster. By combining real-time data during the disaster evolution process, the weights are automatically adjusted according to the stage.

[0055] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. 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 multi-scale resilience assessment method for distribution networks based on multi-hazard coupling modeling, characterized in that, A multi-hazard coupling modeling mechanism is introduced into the resilience assessment of distribution networks. This mechanism comprehensively considers various influencing factors such as typhoons, rainstorms, terrain, and equipment health status. By improving the wind field model, the log-normal rainfall model, and the spatially coupled propagation matrix, a continuous spatiotemporal expression of disaster intensity is achieved. Fuzzy entropy objective weights and Bayesian subjective weights are organically integrated, and a time-varying adjustment factor enables the weights to self-adjust as the disaster situation evolves. This overcomes the static and rigid limitations of traditional combined weighting methods, automatically adjusting indicator weights according to the disaster stage, thus achieving dynamic evolution and intelligent adaptation of the resilience assessment system.

2. A multi-scale resilience assessment method for distribution networks based on multi-hazard coupling modeling, characterized in that, Includes the following steps: S1: Establish a typhoon wind field model for the area where the power distribution network is located. It is reasonably assumed that the typhoon intensity remains constant within this time range. Based on the Batts wind field model, the wind speed at each point within the influence range of the typhoon wind circle is simulated by superimposing the gradient wind speed in the wind circle with the typhoon's moving speed. S2: Establish a typhoon path model for the area where the power distribution network is located. Taking into full account the geographical location of the power distribution network and historical typhoon path information, a combination of probability distribution fitting of historical typhoon path data and Monte Carlo random sampling is adopted to construct a typhoon disaster path model. S3: Distribution network fault scenario simulation based on multi-hazard coupling modeling. It integrates the coupling relationship between typhoon, rainfall, terrain and equipment health status, uses model Carlo sampling to randomly simulate distribution network fault scenarios under extreme events, and uses mathematical statistics to solve the selected distribution network fault characteristics. S4: Load reduction based on load importance. Through load importance analysis of the distribution network and the different power supply reliability requirements of different load levels, the loads at each node are classified into Level 1, Level 2, and Level 3 loads. Load reduction schemes are then determined for each load level. S5: The selection of resilience assessment indicators takes into account the changes in the internal power supply capacity of the distribution network during the occurrence of disasters, as well as the regional characteristics of external disasters. For this type of complex decision-making problem with multiple criteria, the analytic hierarchy process is used to hierarchically divide the assessment indicators and establish a hierarchical resilience assessment indicator system. S6: Calculate objective weights based on fuzzy entropy. In order to fully reflect the objective differences and uncertainties of resilience assessment indicators, this step uses the fuzzy entropy method to quantify the information content of the indicators and calculate the objective weights. The fuzzy entropy method can more accurately characterize the information contribution of the indicators when the sample data has fuzziness, uncertainty or nonlinear characteristics. S7: Adaptive combined weighting and fusion, under complex disaster conditions, dynamically adjusts the importance of each indicator with the changes in time and disaster stage, and combines real-time data in the disaster evolution process to achieve self-adjustment of weights with stage.

3. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: In step S1, the wind field distribution of the power distribution network area is first constructed using the Batts typhoon wind field model. Since the research object is a regional power distribution network, the duration of its impact by the typhoon is relatively limited. Therefore, it can be reasonably assumed that the typhoon intensity remains constant during this period. This wind field model describes the wind speed distribution at various locations within the typhoon's wind circle by superimposing the gradient wind speed within the wind circle with the typhoon's moving speed. Friction effects are not considered in this process. The formula for calculating the gradient wind speed is as follows: In the formula: The maximum gradient wind speed (km / h) within the wind circle; K is an empirical coefficient with a value of 6.72; The central pressure difference (hpa); The radius of the typhoon's maximum wind speed (km); The Coriolis force coefficient for Earth's rotation is taken at a geographical latitude of 30°. Based on this, gradient wind speed is used... and typhoon movement speed By superimposing the data, we can obtain the average maximum wind speed within the typhoon's wind circle at a height of 20m above the sea surface within 10 minutes. The calculation formula is as follows: Furthermore, when any point on the sea surface is a distance *r* from the typhoon center, the typhoon wind speed at that location is: In the formula: Let r be the average wind speed (km / h) at any point in the typhoon wind field at a distance r from the typhoon center. This represents the average maximum wind speed within the typhoon's wind circle. denoted as , where is the radius of the typhoon's maximum wind speed (km); 7r is the distance from the study point to the typhoon's center (km). After the typhoon makes landfall, the reduction effect of terrain roughness (such as mountains, basins, or building complexes) on wind speed needs to be considered. Therefore, a terrain correction coefficient is introduced into the model to correct for wind speed, thus obtaining the formula for calculating the 10-minute average wind speed at a height of 10m on land: In the formula: is the average wind speed (km / h) at any point in the typhoon wind field on land. denoted as the average wind speed (km / h) at any point in the typhoon wind field over the sea; p is the obstacle factor, taken as 0.85; z is the ground roughness length, which is taken as 1 according to the roughness classification standard since the study focuses on urban power distribution networks.

4. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: In step S2, a typhoon path model for the study area needs to be established. This model must reflect both the historical typhoon movement characteristics and the geographical conditions of the target power distribution network location. The specific process is as follows: (1) Set the simulation area: Determine the radius with the research point as the center. The simulated circle is used to determine whether a typhoon will affect the study point. The minimum distance between the study point and the typhoon's path... Smaller than the radius of the simulated circle They believed that the typhoon would affect the research site; otherwise, it would have no impact. (2) Historical data screening and preprocessing: In order to ensure the reasonableness of the probability distribution fitting, only typhoon, strong typhoon and super typhoon events with wind force exceeding level 12 at the time of landfall are retained, while ensuring that the parameter values ​​are within a reasonable range; (3) Probability distribution fitting of key parameters: Statistical modeling and distribution fitting of key parameters such as the typhoon's moving speed, wind direction, and minimum distance from the study point to the typhoon's path. (4) Typhoon path generation: By using Monte Carlo random sampling, a set of parameters (moving speed, wind direction and minimum distance) are extracted from the fitted probability distribution to construct a virtual typhoon path; (5) Multi-path expansion: If multiple virtual typhoon paths need to be generated, the sampling and construction process is repeated to form a set of typhoon paths for multiple scenarios; The typhoon path model established through the above steps can introduce uncertainty modeling while taking into account historical statistical characteristics, laying the foundation for subsequent disaster scenario simulation.

5. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: In step S3, a multi-hazard coupled model considering typhoons, rainstorms, topography, and equipment health status is established. A probabilistic model is constructed to describe the multi-dimensional impact of typhoons and rainstorms on active power distribution networks: typhoons generate high wind speeds, causing mechanical damage to overhead lines and tower structures; rainstorms lead to increased soil moisture, causing changes in grounding resistance and equipment insulation degradation; topography and geomorphological features affect the distribution of wind and rain fields and the path of disaster propagation; equipment aging and health status determine differences in disaster vulnerability, etc. Its parameters represent the operating status and fault probability of distribution network components, and a multi-hazard coupling strength function is introduced. The comprehensive representation of the node at time 1 Intensity of the disaster impact: in, For nodes Wind speed, For nodes At the intensity of rainfall, For nodes Terrain exposure coefficient The device health index ([0,1], the smaller the value, the more vulnerable the device). , , , For disaster factor weighting coefficients, satisfying ; In the disaster modeling process, heavy rain can cause equipment immersion and insulation degradation. A modified log-normal rainfall model is used to describe the rainfall intensity distribution in different regions: in, This represents the maximum rainfall intensity. The distance from the node to the center of the rainstorm; The time-dependent disturbance factor reflects the change in rainfall intensity over time. The insulation degradation rate caused by rainfall is expressed as: in The aging rate of equipment under normal conditions. This is the rainfall sensitivity coefficient.

6. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: An improved equipment cumulative damage model and node failure rate calculation method are adopted to measure equipment health. Reflecting its remaining performance under disaster conditions, a degradation differential equation is defined based on historical equipment operation and maintenance data and disaster stress levels: in This represents the vulnerability coefficient related to the device type. Integral yields: when When this occurs, the equipment is considered to have failed, or the failure rate of the node or line is considered to be high. Comprehensive Disaster Intensity There is a logical relationship: in This is a spatial coupling term used to describe the fault propagation effect between adjacent nodes: in The disaster propagation coefficient, Let be the delay time that a fault in node j affects in node i.

7. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: Disaster scenarios are generated based on the spatiotemporal coupling characteristics of disasters. Integrating the coupling relationship between typhoons, rainfall, terrain, and equipment health status, a Monte Carlo sampling method is used to generate a multi-disaster scenario set. Each scenario includes wind speed, rainfall intensity, failure rate, and equipment status at different time series. The comprehensive disaster intensity distribution of each scenario can be represented as: By generating random numbers using a computer and determining whether a component is faulty based on a probability distribution, the operating state of the power distribution network at a certain moment can be obtained. If the total number of components is N, its operating state can be represented by a vector: Operating status of each component It can be determined by the uniformly distributed random number r and the failure probability p: Repeat the above process M times to obtain M power distribution network system state samples, which represent various fault scenarios that may occur under extreme disasters: Convergence and Accuracy Criteria: According to the law of large numbers, as the number of samplings increases, the simulation results will gradually approach the true values. Combining this with the central limit theorem, the variance coefficient can be used as a convergence criterion. Topology analysis: When sampling results show that a component has failed and is out of service, it is necessary to recalculate the topology of the distribution network to determine whether the connectivity and power flow distribution have changed. The distribution network consists of nodes and switching lines. Graph theory methods can be used to perform topology analysis to verify node reachability and system partitioning. When a system component is found to be faulty during sampling, the component shuts down. This may alter the connectivity of the original distribution network, affecting the power flow distribution. Therefore, topology analysis is needed to determine the number of nodes and their connections after a fault. A distribution network consists of nodes and switching lines. Graph theory can be used for topology analysis. Solving graph theory problems often requires graph search, which involves starting from a vertex and sequentially visiting the remaining vertices, with each vertex visited at most once. The basic idea is to select a vertex, mark it, search its neighbors and mark them, and continue searching the neighbors of those neighbors until all vertices in the graph have been marked.

8. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: In step S4, the concept of load importance is introduced to design load reduction strategies. Different nodes have varying degrees of dependence on system power supply reliability; therefore, they need to be classified according to importance. Based on the classification results, reasonable load reduction schemes are formulated. Specifically, first, the importance of each load in the distribution network is analyzed, and considering the differentiated power supply reliability requirements of different load levels, all node loads are divided into three levels: Level 1, Level 2, and Level 3 loads. Then, based on the classification results, corresponding load reduction strategies are formulated for different load levels. Load classification principles and application scenarios: Based on the requirements for power supply continuity, node loads are divided into three levels: Primary loads require a continuous power supply under all circumstances. Typical application environments include: 1) Places where a power outage would directly cause a life-threatening accident, such as hospitals at level two or above; 2) Industries where power outages would cause serious political or economic losses, such as product processing based on important raw materials, steel smelting, rocket launch sites, and fire protection power supply for high-rise buildings; 3) Key power consumption locations that have a significant impact on the political and economic situation, such as important transportation hubs, star-rated hotels, large conference and exhibition centers, and important public facilities; For primary loads, there are power supply solutions such as dual power supply and dedicated power supply. At the same time, the application of distributed generation provides another convenient means to ensure the continuous power supply of such loads. When the distribution network fails, distributed generation can quickly supply power to primary loads within its capacity. Although this solution is slightly less economical, it can significantly save investment in distribution network equipment, and its overall benefits are still good after optimization. Secondary loads are required to maintain power supply as much as possible, but partial interruptions are permissible in extreme cases. Common scenarios include: 1) Enterprises that would have a significant political or economic impact if a power outage occurs, such as continuous production enterprises; 2) Power outages would disrupt the normal operation of important public facilities, such as transportation hubs, large shopping malls, large stadiums and theaters; For this type of load, dual-circuit power supply and dual-transformer power supply are adopted. Meanwhile, distributed generation can also serve as a supplementary measure. In the event of a fault in the distribution network, distributed generation can provide timely power support to some or even all of the secondary loads, provided capacity allows. Level 3 loads do not have strict requirements for power supply continuity. Typical applications include rural residential electricity consumption and most township enterprise electricity consumption, as well as ordinary loads that do not belong to Level 1 or Level 2 loads. For these types of loads, distributed power sources can be used as random power sources to fully realize their economic, social, and environmental benefits. It should be emphasized that load classification is relative and should be reasonably divided in combination with the local power security level and taking into account political and economic impacts. Since the resilience of a distribution network is often reflected in its ability to support and recover critical loads, different weights should be assigned to different load levels during the assessment: Level 1 loads are assigned a value of 6, Level 2 loads are assigned a value of 3, and Level 3 loads are assigned a value of 1. This differentiated weighting can more intuitively reflect the influence of various types of loads in the resilience assessment. When a system failure occurs and topology analysis reveals line overload, load shedding must be performed to restore the safe and stable operation of the distribution network. During this process, the goal of load shedding should be to minimize the amount of load removed while ensuring system safety, and to prioritize preventing the shedding of critical loads. The specific steps are as follows: 1) In the distribution network topology after the fault, each branch is processed in layers starting from the power source node, and then searched one by one in the order of the layers. 2) When an overload is detected on a line, first determine the magnitude of the overload power; 3) Select appropriate load shedding combinations based on load levels to ensure that the total amount of load shedding is not less than the overload power. When implementing load shedding, follow the principle of "from low to high": first shedding level 3 loads; if this is still insufficient to eliminate the overload, then shedding level 2 loads in sequence, and finally considering level 1 loads. 4) Continue searching the system according to the topology hierarchy. If other lines are found to be overloaded, repeat steps 2) and 3) until all lines in the system are restored to a safe operating state.

9. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: In step S5, a resilience assessment index system is constructed. This system must reflect both the dynamic changes in the internal power supply capacity of the distribution network after a disaster and the regional characteristics of the external disaster. To address this type of complex decision-making problem involving multiple criteria, this invention employs the Analytic Hierarchy Process (AHP) to establish a multi-scale resilience index framework by hierarchically classifying the indicators. The entire index system is divided into three levels: (1) Target layer: that is, assessing the overall resilience level of the distribution network. (2) Primary indicator layer: This layer forms the framework of the resilience assessment system and covers core indicators in different dimensions. (3) Secondary indicator layer: refined to specific measurement parameters Table 1 Resilience Assessment Index Framework The specific contents of this indicator system are as follows: (1) Distribution network power supply capacity: including the following three aspects: missing area of ​​system function curve, disaster absorption rate and post-disaster recovery rate. The missing area of ​​system function curve can be measured by drawing dynamic function curves before, during and after the disaster. The load importance index is defined as the load importance weight at the current time t. With loss of load The product of is calculated using the following formula: Therefore, the missing area of ​​the real-time function curve and the system function curve considering load importance can be expressed as: The disaster absorption rate index of a distribution network represents the ratio of the load level that the system can maintain during a disaster to the initial load level before the disaster. It is used to characterize the distribution network's ability to withstand external shocks. The calculation formula is as follows: In the formula: The load level that can maintain continuous power supply during a disaster; This represents the initial load level of the system. The disaster recovery rate of a distribution network measures how quickly a distribution network can restore its initial power supply level after a disaster. The formula for its calculation is: In the formula: This represents the missing values ​​for the load level of the distribution network during the disaster. The time it takes for a distribution network to recover power to all loads after a load loss; (2) The degree of impact of disasters on the power distribution network is mainly described by the following three aspects: including the frequency of typhoon disasters, the average wind speed of typhoons, and the duration of typhoon impact on the power distribution network; the formula for calculating the duration of typhoon impact T is as follows: In the formula: The radius of the simulated circle in the typhoon model; This is the minimum distance from the geographical center of the power distribution network to the typhoon's path.

10. The method for multi-scale resilience assessment of distribution networks based on multi-hazard coupling modeling according to claim 1, characterized in that: In step S6, in order to fully reflect the objective differences and uncertainties of resilience assessment indicators, this step uses the fuzzy entropy method to quantify the information content of the indicators and calculate the objective weights. Compared to the traditional entropy weight method, the fuzzy entropy method can more accurately characterize the information contribution of indicators when the sample data has fuzziness, uncertainty, or nonlinear characteristics. The specific method is as follows: 1) Constructing a data matrix It has Each evaluation indicator and For each evaluated object, the raw data for each indicator can be represented by a matrix, where each row of the matrix represents the number of evaluated objects. The column represents the number of evaluation indicators. : 2) Data standardization processing For data matrix Standardization is performed to obtain a standardized data matrix. The processing method for each indicator can be selected according to the indicator type. The specific methods are as follows: Profitability metrics: Cost indicators: 3) Membership function calculation: By mapping each standardized index value to the "degree of membership to the ideal state," the uncertainty of the sample is reflected. The membership degree is defined as: Meanwhile, considering the impact of indicator ambiguity, adjustment parameters are introduced. , representing the fuzzy recognition sensitivity of the indicator, the improved membership degree is: in Let be the mean membership degree of the j-th index. This processing helps mitigate the distortion effect of extreme samples on the overall weight distribution; 4) Fuzzy entropy value calculation: Fuzzy entropy value The formula for calculating the j-th index reflects the uncertainty of information. When the indicator samples have large differences and high information content It has a wide distribution in the [0,1] interval, leading to Smaller, when sample differences are small and information content is low. Concentrated distribution Larger; 5) Calculation of objective weights The fuzzy entropy weights are obtained by normalizing the effective information content: This weight reflects the objective contribution of each indicator to changes in system resilience, effectively avoiding the oversensitivity of traditional entropy weight methods to extreme samples. The calculated... This will serve as an objective benchmark for subsequent adaptive combination weighting and fusion; Under complex disaster conditions, the importance of each indicator dynamically adjusts with time and disaster stage. To address this, an adaptive combined weighting algorithm based on fuzzy entropy-Bayes fusion is proposed. By combining real-time data during the disaster evolution process, the weights are automatically adjusted according to the stage, and a multi-scale resilience dynamic assessment framework is constructed. The specific process is as follows: 1) Dynamic update of Bayesian subjective weights: in: Let t be the explanatory probability of the j-th index for changes in system resilience. The subjective weighting of the previous stage; This formula allows the indicator weights to be adaptively updated based on real-time disaster observation data. For example, if changes in the system function curve recovery rate during the disaster phase significantly affect the resilience results, then the weights of this indicator... Increase Automatic ascent reflects a dynamic subjective correction mechanism; 2) Adaptive weight fusion: To comprehensively consider both the fuzzy entropy weights of objective data and the subjective Bayesian weights, a time-varying adjustment factor is defined. : in: Let be the standard deviation of the fluctuation of indicator j within the time window; This represents the average standard deviation of the volatility of all indicators; when the indicators are highly volatile (information changes rapidly), When the value increases, the model tends to use objective entropy weights; when the indicator changes slowly or is dominated by experience, the model tends to use objective entropy weights. The reduction relies more on subjective Bayesian weights; 3) Calculation of combined weights: The final adaptive combined weights are defined as follows: The beneficial effects of this invention are as follows: Compared with existing technologies, the multi-scale resilience assessment method for distribution networks based on multi-hazard coupling modeling proposed in this invention has the following advantages and positive effects: This invention introduces a multi-hazard coupled modeling mechanism in distribution network resilience assessment, comprehensively considering multiple influencing factors such as typhoons, rainstorms, topography, and equipment health status. Through improved wind field models, log-normal rainfall models, and the linked modeling of spatially coupled propagation matrices, it achieves a continuous spatiotemporal expression of hazard intensity. This model not only reflects the local destructive effects of a single hazard but also reveals the evolution of node failure probability and the propagation characteristics of regional vulnerability under the combined impact of multiple hazards. The fuzzy entropy method is used to calculate the information content of the indicators, and the introduction of membership functions and fuzzy recognition factors ensures computational stability even when data exhibits fuzziness, uncertainty, or nonlinear distribution. This method can more accurately characterize the information contribution of different resilience indicators under disaster conditions, significantly improve the robustness and interpretability of weight calculation, and provide a solid objective foundation for indicator evaluation in complex disaster scenarios. Simultaneously, it organically integrates fuzzy entropy objective weights with Bayesian subjective weights, achieving self-adjustment of weights as the disaster situation evolves through time-varying adjustment factors. It overcomes the static and rigid limitations of traditional combined weighting methods, enabling automatic adjustment of indicator weights according to the disaster stage, thus achieving dynamic evolution and intelligent adaptation of the resilience assessment system. In summary, this invention can significantly improve the risk resistance and recovery level of power distribution networks under extreme disaster conditions while ensuring technical feasibility, and has high practical value and application prospects.

Images