Multi-hazard exposure driven power grid risk resilience collaborative optimization method

Abstract

The present application relates to the technical field of power grid risk assessment and resilience optimization, and discloses a multi-disaster exposure driven power grid risk resilience collaborative optimization method, which obtains extreme weather, power grid topology and line parameters, distributed energy, load, electric vehicles and energy storage and historical fault data, and constructs an operating environment state vector; combines multiple meteorological factors to calculate a fault rate increment, introduces exposure factors such as line length and terrain exposure degree, differentially corrects a line unit length composite fault rate, obtains a line fault probability, establishes a weather variable probability model, generates an extreme weather scenario and matches source load storage operating states, and constructs a joint operating scenario; quantifies multi-dimensional fault severity through fault flow calculation, calculates system expected comprehensive risk, constructs a risk-cost-resilience three-target collaborative optimization model and solves an output scheme. The present application can improve the accuracy of power grid risk assessment under extreme weather and effectively enhance the disaster resistance resilience of the power grid.

Description

Technical Field

[0001] This invention relates to the field of power system collaborative optimization technology, and more specifically, to a collaborative optimization method for power grid risk resilience driven by multi-hazard exposure. Background Technology

[0002] With the increasing integration of distributed photovoltaic, wind power, electric vehicles, and energy storage devices into the power grid, the grid operation is characterized by multiple power sources, multiple loads, and multiple states coupled together. Under normal weather conditions, this type of grid can improve power supply support capacity and operational safety margin. However, under extreme weather conditions such as strong winds, lightning, heavy rain, high temperatures, and low temperatures, its operational risks are significantly complex and sudden: strong winds may increase the mechanical stress on overhead lines and raise the failure rate; lightning may cause line insulation breakdown and equipment tripping; and heavy rain and extreme temperatures may also affect the output of distributed energy resources, the charging and discharging behavior of electric vehicles, the availability of energy storage, and the level of load demand.

[0003] In existing technologies, power grid system reliability assessments typically employ indicators such as average outage duration, average outage frequency, and expected power shortage. While these indicators are suitable for reflecting long-term average reliability levels, they are insufficient for characterizing short-term high-risk conditions caused by extreme weather. Probabilistic risk assessment methods can simultaneously consider the probability of event occurrence and the severity of consequences. Existing research has further combined Monte Carlo simulation, severity functions, line fault probability models, and multi-objective optimization methods to assess the risks of distributed energy systems under extreme weather conditions and reduce system risks through distributed energy capacity configuration.

[0004] However, in practice, the aforementioned existing technologies have been found to have the following technical problems: In traditional risk assessments, the line failure rate is often taken as a constant, or a uniform overall correction is made for each line based solely on weather type. This makes it difficult to reflect the dynamic amplification effect of extreme weather such as strong winds and lightning on the line failure rate, nor can it reflect the differences between different lines in terms of line length, overhead ratio, terrain exposure, and historical failure levels. In reality, during the same strong wind, the failure probability of long lines erected in open terrain with a high proportion of exposed ground differs significantly from that of short lines with better laying conditions. The overall correction method masks this difference, leading to distorted risk assessment results and consequently undermining the reliable basis for subsequent risk-based decisions.

[0005] Existing optimization methods typically compromise between operational risk and economic cost, failing to reflect the system's need to maintain power supply, minimize outages, and improve recovery capabilities after disturbances during extreme weather events. This results in optimizations that, while economically feasible, lack the ability to maintain and rapidly restore power supply when extreme weather actually occurs. Therefore, there is an urgent need for a multi-hazard exposure-driven collaborative optimization method for grid risk resilience. Summary of the Invention

[0006] To address the aforementioned technical issues, this invention provides a collaborative optimization method for power grid risk resilience driven by multi-hazard exposure. This method differentiates and corrects line failure rates, constructs a weather-source-load-storage joint operation scenario, establishes a closed-loop optimization system with three objectives: risk, cost, and resilience, identifies key components and prioritizes their reinforcement, and combines online parameter rolling correction to effectively improve the accuracy of risk assessment and the disaster resistance resilience of the power grid under extreme weather conditions.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows:

[0008] A multi-hazard exposure-driven collaborative optimization method for power grid risk resilience includes:

[0009] Acquire extreme weather data, power grid topology and line parameters, distributed energy and load data, electric vehicle and energy storage data, and historical fault records to construct an operating environment state vector; the extreme weather data includes wind speed, solar irradiance, lightning flash density, ambient temperature, and rainfall intensity.

[0010] The failure rate increment is determined based on wind speed, lightning flash density, rainfall intensity, and ambient temperature. The line exposure factor is determined based on line length, terrain exposure degree, overhead ratio, and historical failure count. The line exposure factor is used to correct the composite failure rate per unit length of the line to obtain the line failure probability.

[0011] Establish probabilistic models for various weather variables, generate extreme weather scenarios, and combine these extreme weather scenarios to determine the scenario-based operating status of distributed energy, electric vehicles, energy storage, and loads, thereby constructing joint operation scenarios.

[0012] In each joint operation scenario, the grid topology is updated and AC power flow calculations are performed for line fault events to obtain node voltage, line load rate, unsupplied load, and information on islanded or power-out nodes;

[0013] Based on the node voltage, line load rate, unsupplied load, and islanded or power-out node information, determine the severity of node low voltage, line overload, load loss, and islanding impact. Calculate the comprehensive consequence severity corresponding to the line fault event based on the four severity levels determined. Combine the line fault probability with the corresponding comprehensive consequence severity to calculate the expected comprehensive risk of the system.

[0014] Based on the expected comprehensive risk of the system, risk contribution is decomposed to identify key components with risk contribution rates higher than a threshold, calculate the unit reinforcement benefit and determine the reinforcement priority; the key components and their reinforcement priorities are used as inputs for the key component reinforcement decision.

[0015] With the expected comprehensive risk, total cost, and system resilience as optimization objectives, a three-objective collaborative optimization model is established, which includes distributed energy capacity allocation and key component reinforcement decisions.

[0016] Solving the three-objective collaborative optimization model yields the distributed energy capacity configuration results for each node, the key component reinforcement decision results, the key component reinforcement priority, and the corresponding expected comprehensive risk, total cost, and system resilience index values. These are then used as the output of the power grid collaborative optimization scheme.

[0017] As a preferred embodiment of the present invention, the failure rate increment includes the failure rate increment due to strong winds, lightning, heavy rain, and high or low temperature triggering the failure rate. Specifically, the failure rate increment due to strong winds is zero when the wind speed does not exceed the critical wind speed, and increases with the magnitude of the wind speed exceeding the critical wind speed. The failure rate increment due to lightning is zero when the lightning flash density does not exceed the threshold, and increases with the magnitude of the lightning flash density exceeding the threshold. The failure rate increments due to heavy rain, high or low temperature are determined based on the disaster triggering threshold, intensity segmentation coefficient, duration, and equipment tolerance level.

[0018] As a preferred embodiment of the present invention, the line exposure factor is determined by weighting the normalized results of line length, terrain exposure degree, overhead ratio and number of historical faults, and the sum of the weights of each exposure factor is 1.

[0019] As a preferred embodiment of the present invention, the generation of extreme weather scenarios specifically includes: establishing probability models for wind speed, solar irradiance, lightning variables, ambient temperature and rainfall intensity respectively, and performing scenario sampling or clustering based on historical weather records of the same period to obtain a set of extreme weather scenarios including wind speed, solar irradiance, lightning ground flash density, ambient temperature and rainfall intensity, and determining the probability of occurrence of each extreme weather scenario.

[0020] As a preferred embodiment of the present invention, the specific method for determining the scenario-based operating state in conjunction with extreme weather scenarios includes:

[0021] Based on the solar irradiance and ambient temperature of the corresponding scenario, the photovoltaic output is calculated through photovoltaic cell temperature, output current, voltage, and fill factor; based on the wind speed of the corresponding scenario, the wind power output is determined according to the segmented relationship of cut-in wind speed, rated wind speed, and cut-out wind speed; the aggregate power of electric vehicles is calculated based on the number of electric vehicles connected, the charging and discharging probability during the time period, and the charging and discharging power of a single vehicle; a net load following strategy is adopted, and the energy storage charging and discharging power and time-series energy status are determined by combining the node net load status and the upper and lower limits of energy storage; the node load is sampled using a normal distribution random model to obtain the scenario load value, thereby realizing the synchronous mapping of weather disturbances to the source, load, and storage operation variables.

[0022] As a preferred embodiment of the present invention, the expected comprehensive risk of the calculation system includes: in each joint operation scenario and each evaluation period, for each line fault event in the set of line fault events, setting the corresponding line to a fault state and updating the power grid topology, performing AC power flow calculation, and obtaining the comprehensive consequence severity of the line fault event based on the severity of node low voltage, line overload, load loss, and islanding impact; combining the line fault probability of the line fault event with the corresponding comprehensive consequence severity to obtain the risk value of the line fault event; summarizing the risk values ​​of each line fault event in the set of line fault events to obtain the scenario comprehensive risk under the corresponding joint operation scenario and evaluation period; and summarizing the scenario comprehensive risks under each joint operation scenario and each evaluation period to obtain the expected comprehensive risk of the system.

[0023] As a preferred embodiment of the present invention, the identification of key components with risk contribution rates exceeding a threshold, calculation of unit reinforcement benefits, and determination of reinforcement priorities include: for line components, determining their risk contribution value based on the expected risk generated by the corresponding line fault event in all joint operation scenarios and evaluation periods; for nodes, distributed energy, energy storage, and electric vehicle resources, using a counterfactual marginal risk contribution method, comparing the expected comprehensive risk of the corresponding resource under normal availability with the expected comprehensive risk of the corresponding resource under reduced capacity, shutdown, or reduced availability, and determining their risk contribution value based on the difference; determining the corresponding risk contribution rate based on the proportion of each component or resource's risk contribution value to the total risk contribution value of all components or resources participating in risk ranking; screening components or resources with risk contribution rates exceeding a preset threshold as key components; for each key component, determining the unit reinforcement benefit based on the reduction between the expected comprehensive risk of the system before reinforcement and the expected comprehensive risk of the system after reinforcement, as well as the reinforcement cost of the key component, and determining the reinforcement priority of key components according to the unit reinforcement benefit from high to low.

[0024] As a preferred embodiment of the present invention, in the three-objective collaborative optimization model, the total cost consists of three parts: the investment cost of distributed energy capacity configuration and key component reinforcement, the operation and maintenance cost of distributed energy, and the expected power outage loss cost; the system resilience index is represented by the ratio of the integral of the actual load supply ratio of the system during the period of extreme weather impact to the integral of the load supply ratio under undisturbed conditions.

[0025] As a preferred embodiment of the present invention, it also includes online rolling correction, specifically: after each extreme weather process, the actual line fault observation results, operation status data and model calculation results are compared, the mean square error of the fault observation values ​​and the fault probability calculated by the model within the historical period is used as the loss function, and the fault rate parameters corresponding to strong wind, lightning, rainstorm, high temperature and low temperature are iteratively updated by using gradient descent, while the severity weight and weather scenario probability are corrected simultaneously.

[0026] This invention also provides a multi-hazard exposure-driven power grid risk resilience collaborative optimization system, comprising:

[0027] The operating environment perception module is used to acquire extreme weather data, power grid topology and line parameters, distributed energy and load data, electric vehicle and energy storage data, and historical fault records to construct an operating environment state vector; the extreme weather data includes wind speed, solar irradiance, lightning flash density, ambient temperature, and rainfall intensity.

[0028] The line fault probability calculation module is used to determine the fault rate increment based on wind speed, lightning flash density, rainfall intensity, and ambient temperature, and to determine the line exposure factor based on line length, terrain exposure degree, overhead ratio, and historical fault count. The line exposure factor is used to correct the composite fault rate per unit length of the line to obtain the line fault probability.

[0029] The joint operation scenario generation module is used to establish probabilistic models of various weather variables, generate extreme weather scenarios, and combine extreme weather scenarios to determine the scenario-based operation status of distributed energy, electric vehicles, energy storage and load, and construct joint operation scenarios.

[0030] The fault scenario power flow calculation module is used to update the power grid topology and perform AC power flow calculations for line fault events under various joint operation scenarios to obtain node voltage, line load rate, unsupplied load and islanded or power-out node information.

[0031] The system risk quantification module is used to determine the severity of node low voltage, line overload, load loss, and islanding or power failure based on the node voltage, line load rate, unsupplied load, and islanding impact information. Based on the determined four severity levels, it calculates the comprehensive consequence severity corresponding to the line fault event and combines the line fault probability with the corresponding comprehensive consequence severity to calculate the expected comprehensive risk of the system.

[0032] The critical component identification module is used to decompose the risk contribution based on the expected comprehensive risk of the system, identify critical components with a risk contribution rate higher than a threshold, calculate the unit reinforcement benefit and determine the reinforcement priority; and use the critical components and their reinforcement priorities as inputs to the critical component reinforcement decision.

[0033] The collaborative optimization model construction module is used to establish a three-objective collaborative optimization model with the system's expected comprehensive risk, total cost, and system resilience as optimization objectives, including distributed energy capacity configuration and key component reinforcement decisions.

[0034] The optimization solution and output module is used to solve the three-objective collaborative optimization model, obtain the distributed energy capacity configuration results of each node, the key component reinforcement decision results, the key component reinforcement priority, and the corresponding expected comprehensive risk, total cost and system resilience index values, and output them as the power grid collaborative optimization scheme.

[0035] The beneficial technical effects of this invention are:

[0036] This method constructs incremental failure rate models for multiple disaster types, including strong winds, lightning, rainstorms, high temperatures, and low temperatures. It also introduces a line exposure factor that includes line length, terrain exposure level, overhead ratio, and historical failure level to differentiate the composite failure rate of each line. This method can truly reflect the significant differences in the probability of line failure under different laying conditions and geographical environments, avoiding the risk assessment distortion caused by overall correction, and providing a reliable data foundation for subsequent risk calculation and optimization decisions.

[0037] It generates extreme weather scenarios that maintain the intrinsic correlation of weather variables, and simultaneously maps weather disturbances to line fault probability, distributed energy output, electric vehicle charging and discharging behavior, energy storage operation status and node load demand. It constructs a weather-equipment-load joint operation scenario, which fully covers the multiple impact paths of extreme weather on the power grid system and is more in line with the complex and sudden characteristics of risks in actual operation.

[0038] In addition to system risk and total cost, system resilience is taken as an independent optimization objective. At the same time, a composite severity evaluation system covering node low voltage, line overload, load loss and islanding impact is established. Through hardening decisions, the correction of failure rate, capacity parameters and resource availability is fed back to risk calculation to form an optimization closed loop. The three are balanced and optimized in the distributed energy capacity configuration and key component hardening scheme, so that the scheme can effectively improve the system's power supply maintenance capability and fault recovery capability under extreme weather conditions while meeting economic requirements.

[0039] By accurately identifying the key components that have the greatest impact on system risk through risk contribution rate, and prioritizing reinforcement based on unit reinforcement benefit, we can provide operation and maintenance and planning personnel with a clear risk handling sequence and resource allocation direction, so that the limited reinforcement budget can achieve the maximum risk reduction effect, which has both theoretical optimization value and engineering practicality.

[0040] An online rolling correction mechanism is set up. Based on actual fault observation and operation data, the fault rate parameter, severity weight and weather scenario probability are updated iteratively through gradient descent using mean square error as the loss function. It can be adapted to power grid systems with different regions, seasons and grid structures. As operation data accumulates, the accuracy of risk assessment and collaborative optimization is continuously improved. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the overall process of the present invention.

[0042] Figure 2 It is a curve showing the change in line failure rate under extreme weather conditions.

[0043] Figure 3 This is a schematic diagram illustrating the coupling mechanism of extreme weather, equipment failure, and system risk.

[0044] Figure 4 These are schematic diagrams of the probability distribution of wind speed and solar irradiance and typical scenarios; where (a) is a schematic diagram of wind speed and typical scenarios, and (b) is a schematic diagram of the probability distribution of solar irradiance and typical scenarios.

[0045] Figure 5 This is a schematic diagram of typical power curves for distributed energy, energy storage, and load.

[0046] Figure 6 It is a schematic diagram of a power grid system structure that includes distributed energy resources, electric vehicles, and energy storage.

[0047] Figure 7 These are schematic diagrams of the assessment results for the severity of node undervoltage and line overload; where (a) is a schematic diagram of the assessment results for the severity of node undervoltage and (b) is a schematic diagram of the assessment results for the severity of line overload.

[0048] Figure 8 The diagrams are comparative diagrams of comprehensive risk assessment results and methods; (a) is a comparative diagram of comprehensive risk values, (b) is a comparative diagram of risk costs, (c) is a comparative diagram of load loss, and (d) is a comparative diagram of resilience indicators.

[0049] Figure 9 The diagram shows the risk contribution rate of key components and the comparison before and after optimization; (a) is a diagram showing the risk contribution rate of key components, and (b) is a diagram showing the comparison between the overall risk and the expected power outage cost before and after optimization.

[0050] Figure 10 This is a schematic diagram of the optimization results for the three objectives of risk, cost, and resilience. Detailed Implementation

[0051] In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0052] Combination Figures 1-10 The present invention provides the following embodiments:

[0053] Example 1:

[0054] A multi-hazard exposure-driven collaborative optimization method for power grid risk resilience, characterized by comprising:

[0055] Acquire extreme weather data, power grid topology and line parameters, distributed energy and load data, electric vehicle and energy storage data, and historical fault records to construct an operating environment state vector; the extreme weather data includes wind speed, solar irradiance, lightning flash density, ambient temperature, and rainfall intensity.

[0056] The failure rate increment is determined based on wind speed, lightning flash density, rainfall intensity, and ambient temperature. The line exposure factor is determined based on line length, terrain exposure degree, overhead ratio, and historical failure count. The line exposure factor is used to correct the composite failure rate per unit length of the line to obtain the line failure probability.

[0057] Establish probabilistic models for various weather variables, generate extreme weather scenarios, and combine these extreme weather scenarios to determine the scenario-based operating status of distributed energy, electric vehicles, energy storage, and loads, thereby constructing joint operation scenarios.

[0058] In each joint operation scenario, the grid topology is updated and AC power flow calculations are performed for line fault events to obtain node voltage, line load rate, unsupplied load, and information on islanded or power-out nodes;

[0059] Based on the node voltage, line load rate, unsupplied load, and islanded or power-out node information, determine the severity of node low voltage, line overload, load loss, and islanding impact. Calculate the comprehensive consequence severity corresponding to the line fault event based on the four severity levels determined. Combine the line fault probability with the corresponding comprehensive consequence severity to calculate the expected comprehensive risk of the system.

[0060] Based on the expected comprehensive risk of the system, risk contribution is decomposed to identify key components with risk contribution rates higher than a threshold, calculate the unit reinforcement benefit and determine the reinforcement priority; the key components and their reinforcement priorities are used as inputs for the key component reinforcement decision.

[0061] With the expected comprehensive risk, total cost, and system resilience as optimization objectives, a three-objective collaborative optimization model is established, which includes distributed energy capacity allocation and key component reinforcement decisions.

[0062] Solving the three-objective collaborative optimization model yields the distributed energy capacity configuration results for each node, the key component reinforcement decision results, the key component reinforcement priority, and the corresponding expected comprehensive risk, total cost, and system resilience index values. These are then used as the output of the power grid collaborative optimization scheme.

[0063] Furthermore, the failure rate increment includes the failure rate increment for strong winds, lightning, heavy rain, and high or low temperature failures triggered by ambient temperature. Specifically, the strong wind failure rate increment is zero when the wind speed does not exceed the critical wind speed, and increases with the magnitude of the wind speed exceeding the limit after the wind speed exceeds the critical wind speed. The lightning failure rate increment is zero when the lightning flash density does not exceed the threshold, and increases with the magnitude of the lightning flash density exceeding the limit after the lightning flash density exceeds the threshold. The failure rate increments for heavy rain, high or low temperature are determined based on the disaster trigger threshold, intensity segmentation coefficient, duration, and equipment tolerance level.

[0064] The physical mechanisms that cause line failures differ depending on the type of disaster. Strong winds mainly cause line failures through the accumulation of mechanical loads, lightning mainly causes failures through insulation impact breakdown, and rainstorms and extreme temperatures induce failures through equipment exceeding environmental tolerance limits. By setting threshold triggering rules and differentiated growth patterns for each type of disaster, the disaster-causing patterns of various extreme weather events can be accurately matched, avoiding the problem of insufficient adaptability of a single correction model. This makes the increase in failure rate more consistent with the actual fault characteristics of power grid lines as weather intensity changes.

[0065] Furthermore, the line exposure factor is determined by weighting the normalized results of line length, terrain exposure level, overhead ratio, and number of historical faults, with the sum of the weights of each exposure factor being 1.

[0066] The degree of exposure of a line to extreme weather is determined by its own physical properties, laying conditions, geographical environment and historical operating characteristics. By normalizing and weighting multi-dimensional indicators, influencing factors with different dimensions can be integrated into a unified exposure factor quantification value. This objectively distinguishes the differences in disaster resistance capabilities between long and short lines, overhead and cable lines, and lines in open and concealed terrain. It solves the problem of uniform fault rate correction for the entire network masking individual differences of the lines, making the corrected fault rate more consistent with the actual disaster risk of each line.

[0067] Furthermore, the generation of extreme weather scenarios specifically includes: establishing probability models for wind speed, solar irradiance, lightning variables, ambient temperature, and rainfall intensity, and performing scenario sampling or clustering based on historical weather records for the same period to obtain a set of extreme weather scenarios including wind speed, solar irradiance, lightning ground flash density, ambient temperature, and rainfall intensity, and determining the probability of occurrence of each extreme weather scenario.

[0068] In real meteorological environments, weather elements such as wind speed, solar irradiance, lightning, temperature, and rainfall have natural interrelationships. Independent random sampling can easily generate scenario combinations that do not conform to meteorological laws. Sampling or clustering based on historical weather records of the same period can preserve the inherent correlation between multiple types of weather variables, ensuring that the generated extreme weather scenarios conform to real meteorological characteristics, thereby improving the credibility and accuracy of subsequent joint operation scenarios and risk assessment results.

[0069] Furthermore, specific methods for determining scenario-based operational status in conjunction with extreme weather scenarios include:

[0070] Based on the solar irradiance and ambient temperature of the corresponding scenario, the photovoltaic output is calculated through photovoltaic cell temperature, output current, voltage, and fill factor; based on the wind speed of the corresponding scenario, the wind power output is determined according to the segmented relationship of cut-in wind speed, rated wind speed, and cut-out wind speed; the aggregate power of electric vehicles is calculated based on the number of electric vehicles connected, the charging and discharging probability during the time period, and the charging and discharging power of a single vehicle; a net load following strategy is adopted, and the energy storage charging and discharging power and time-series energy status are determined by combining the node net load status and the upper and lower limits of energy storage; the node load is sampled using a normal distribution random model to obtain the scenario load value, thereby realizing the synchronous mapping of weather disturbances to the source, load, and storage operation variables.

[0071] By mapping solar irradiance, ambient temperature, and wind speed in extreme weather scenarios to photovoltaic and wind power outputs, and combining the number of electric vehicles connected and their charging and discharging probabilities, energy storage net load following strategies, and node load stochastic models, the source, load, and storage operating states corresponding to each extreme weather scenario can be formed. This allows weather scenarios, distributed energy output, electric vehicle aggregated power, energy storage operating states, and node loads to correspond to each other in the same joint operating scenario, providing consistent scenario-based operating data for subsequent power flow calculations and risk assessments.

[0072] Furthermore, the expected comprehensive risk of the calculation system includes: in each joint operation scenario and each evaluation period, for each line fault event in the set of line fault events, setting the corresponding line to a fault state and updating the power grid topology, performing AC power flow calculation, and obtaining the comprehensive consequence severity of the line fault event based on the severity of node low voltage, line overload, load loss, and islanding impact; combining the line fault probability of the line fault event with the corresponding comprehensive consequence severity to obtain the risk value of the line fault event; summarizing the risk values ​​of each line fault event in the set of line fault events to obtain the scenario comprehensive risk under the corresponding joint operation scenario and evaluation period; and summarizing the scenario comprehensive risks under each joint operation scenario and each evaluation period to obtain the expected comprehensive risk of the system.

[0073] The essence of system comprehensive risk is the coupling of the probability of occurrence of various failure events and the severity of their consequences. By updating the topology and calculating the severity of consequences for each line failure event separately, the actual operational impact caused by each line failure can be accurately matched, avoiding risk attribution bias caused by the post-event allocation of the overall severity of the scenario. Through multi-level calculations of failure event aggregation, scenario weighting, and time period weighting, the risk levels of different extreme weather scenarios and different time periods can be fully covered, and the obtained expected comprehensive risk of the system is more statistically representative and can serve as a reliable basis for risk warning and optimization decision-making.

[0074] Furthermore, the process of identifying key components with risk contribution rates exceeding a threshold, calculating unit reinforcement benefits, and determining reinforcement priorities includes: for line components, determining their risk contribution value based on the expected risk generated by the corresponding line fault event across all joint operation scenarios and evaluation periods; for nodes, distributed energy resources, energy storage, and electric vehicle resources, using a counterfactual marginal risk contribution method, comparing the expected comprehensive risk of the corresponding resource under normal availability with the expected comprehensive risk of the corresponding resource under reduced capacity, out of operation, or reduced availability, and determining its risk contribution value based on the difference; determining the corresponding risk contribution rate based on the proportion of each component or resource's risk contribution value to the total risk contribution value of all components or resources participating in risk ranking; screening components or resources with risk contribution rates exceeding a preset threshold as key components; for each key component, determining the unit reinforcement benefit based on the reduction between the expected comprehensive risk of the system before reinforcement and the expected comprehensive risk of the system after reinforcement, as well as the reinforcement cost of the key component, and determining the reinforcement priority of key components according to the unit reinforcement benefit from high to low.

[0075] Line components and non-line objects such as nodes, distributed energy resources, energy storage, and electric vehicles affect system risk in different ways. For line components, the risk contribution can be directly determined based on the expected risk generated by their corresponding line failure events in various joint operation scenarios and assessment periods. For non-line objects, by comparing the expected comprehensive system risk under normal availability conditions with those under conditions of reduced capacity, shutdown, or reduced availability, the marginal impact of the object on system risk can be quantified in a counterfactual manner. Further screening of key components based on risk contribution rates, and determining the unit reinforcement benefit by combining risk reduction and reinforcement cost, can establish a correlation between the degree of risk contribution and the economics of reinforcement, providing a basis for prioritizing the reinforcement of key components.

[0076] Furthermore, in the three-objective collaborative optimization model, the total cost consists of three parts: the investment cost of distributed energy capacity configuration and key component reinforcement, the operation and maintenance cost of distributed energy, and the expected power outage loss cost; the system resilience index is represented by the ratio of the integral of the actual load supply ratio of the system during the period of extreme weather impact to the integral of the load supply ratio under undisturbed conditions.

[0077] By including the investment costs of distributed energy capacity configuration and reinforcement of key components, the operation and maintenance costs of distributed energy, and the expected power outage loss costs in the total cost, the construction investment, operating expenses, and failure losses of the scheme can be reflected simultaneously. By using the ratio of the integral of the actual load supply ratio during the period of extreme weather impact to the integral of the load supply ratio under undisturbed conditions to characterize the system resilience, the system's ability to maintain load supply during disturbances can be quantified, thus providing a unified evaluation basis for the coordinated trade-off between the three objectives of expected comprehensive risk, total cost, and system resilience.

[0078] Furthermore, it also includes online rolling correction, specifically: after each extreme weather event, the actual line fault observation results, operation status data and model calculation results are compared. The mean square error between the fault observation values ​​and the fault probability calculated by the model within the historical period is used as the loss function. The fault rate parameters corresponding to strong wind, lightning, rainstorm, high temperature and low temperature are iteratively updated using gradient descent. At the same time, the severity weight and weather scenario probability are corrected simultaneously.

[0079] The catastrophic characteristics of extreme weather and the importance of different operational consequences vary with region, season, and grid status. Using fixed failure rate parameters, severity weights, and weather scenario probabilities is difficult to adapt to actual operating conditions in the long term. After each extreme weather event, the actual line fault observation results and operating status data are compared with the model calculation results. The mean square error between the fault observation value and the model fault probability is used as the loss function to iteratively correct the failure rate parameters corresponding to each type of disaster. The severity weights and weather scenario probabilities are also adjusted synchronously based on the actual operating results. This allows the fault probability calculation, consequence evaluation, and scenario probability representation to be continuously updated with actual operating data, thereby improving the adaptability of subsequent risk assessment and collaborative optimization results.

[0080] Example 2:

[0081] A multi-hazard exposure-driven power grid risk resilience collaborative optimization system, characterized by comprising:

[0082] The operating environment perception module is used to acquire extreme weather data, power grid topology and line parameters, distributed energy and load data, electric vehicle and energy storage data, and historical fault records to construct an operating environment state vector; the extreme weather data includes wind speed, solar irradiance, lightning flash density, ambient temperature, and rainfall intensity.

[0083] The line fault probability calculation module is used to determine the fault rate increment based on wind speed, lightning flash density, rainfall intensity, and ambient temperature, and to determine the line exposure factor based on line length, terrain exposure degree, overhead ratio, and historical fault count. The line exposure factor is used to correct the composite fault rate per unit length of the line to obtain the line fault probability.

[0084] The joint operation scenario generation module is used to establish probabilistic models of various weather variables, generate extreme weather scenarios, and combine extreme weather scenarios to determine the scenario-based operation status of distributed energy, electric vehicles, energy storage and load, and construct joint operation scenarios.

[0085] The fault scenario power flow calculation module is used to update the power grid topology and perform AC power flow calculations for line fault events under various joint operation scenarios to obtain node voltage, line load rate, unsupplied load and islanded or power-out node information.

[0086] The system risk quantification module is used to determine the severity of node low voltage, line overload, load loss, and islanding or power failure based on the node voltage, line load rate, unsupplied load, and islanding impact information. Based on the determined four severity levels, it calculates the comprehensive consequence severity corresponding to the line fault event and combines the line fault probability with the corresponding comprehensive consequence severity to calculate the expected comprehensive risk of the system.

[0087] The critical component identification module is used to decompose the risk contribution based on the expected comprehensive risk of the system, identify critical components with a risk contribution rate higher than a threshold, calculate the unit reinforcement benefit and determine the reinforcement priority; and use the critical components and their reinforcement priorities as inputs to the critical component reinforcement decision.

[0088] The collaborative optimization model construction module is used to establish a three-objective collaborative optimization model with the system's expected comprehensive risk, total cost, and system resilience as optimization objectives, including distributed energy capacity configuration and key component reinforcement decisions.

[0089] The optimization solution and output module is used to solve the three-objective collaborative optimization model, obtain the distributed energy capacity configuration results of each node, the key component reinforcement decision results, the key component reinforcement priority, and the corresponding expected comprehensive risk, total cost and system resilience index values, and output them as the power grid collaborative optimization scheme.

[0090] Application example:

[0091] A collaborative optimization method for power grid risk resilience driven by multi-hazard exposure includes the following steps.

[0092] S1. Basic Data Acquisition and Operational Environment State Vector Construction

[0093] Acquire extreme weather data, power grid topology and line parameters, distributed energy capacity, electric vehicle charging and discharging parameters, energy storage device parameters, and node loads. And historical weather and line fault records.

[0094] Extreme weather data includes wind speed Lightning and ground flash density Solar irradiance Ambient temperature Rainfall intensity Power grid topology and line parameters include line length. Line capacity limit Node admittance matrix conductance With susceptance .

[0095] Build runtime environment state vector:

[0096]

[0097] In the formula, It indicates the operating status of distributed energy resources, including the rated capacity, availability, and output status of photovoltaic and wind power. It indicates the operating status of electric vehicles, including the number of electric vehicles connected, the probability of charging and discharging, and the charging and discharging power; It indicates the operating status of the energy storage device, including the current energy of the stored energy, charging and discharging power, charging and discharging efficiency, and the upper and lower limits of the stored energy.

[0098] For the first in the power grid For each route, based on its geospatial coordinates and the route corridor range, the meteorological data of the entire region is mapped to time-varying meteorological parameters at the route level. The mapping rules are as follows:

[0099] For continuous meteorological elements such as wind speed, ambient temperature, rainfall intensity, and solar irradiance, a meteorological grid matching method is employed. Gridded meteorological forecast / observation data for the assessment area are acquired, and the first... The arithmetic mean of the corresponding meteorological values ​​of all meteorological grids within the coverage area of ​​each line corridor is taken as the line-level meteorological parameter for that line, and is denoted as: Line Wind speed at any moment , No. Line ambient temperature at any time , No. Line Rainfall intensity at any time , No. Line Solar irradiance at any time .

[0100] Alternatively, the inverse distance weighted interpolation method can be used to interpolate meteorological parameters corresponding to the spatial location of the line based on observation data from meteorological stations within the evaluation area.

[0101] Lightning-to-ground flash density was determined using a statistical method based on lightning location data. Statistical analysis was performed using lightning-to-ground flash observation data from the lightning location system. Within a certain time The line-level lightning density of a line is calculated by dividing the number of ground flashes within a pre-defined buffer zone (1-3 km on each side of the line) by the area of ​​the buffer zone. .

[0102] S2, Extreme Weather-Driven Line-Level Failure Rate Correction

[0103] The purpose of this step is to convert weather intensity into line-level fault probability. Unlike a uniform correction of the entire network fault rate based solely on weather type, this method first calculates the base fault rate increment caused by multiple weather hazards, then combines this with factors such as line length, overhead line ratio, terrain exposure level, and historical fault levels to form a line-level exposure correction, ensuring that different lines have different fault probabilities under the same extreme weather conditions. The line-level meteorological parameters corresponding to each of the aforementioned hazards are all obtained through the spatial mapping method described above, thereby achieving differentiated calculation of fault rates for different lines under the same extreme weather conditions. The specific details of this step are as follows.

[0104] S2.1, Constructing the first The line is in Composite failure rate at any time

[0105] No. The line is in The composite failure rate per unit length at time t is:

[0106]

[0107] In the formula, The basic failure rate per unit length under normal weather conditions, its initial value can be calculated as follows: Sure, For the first time under normal weather conditions within the statistical period The number of faults on each line; This refers to the line length; For statistical duration; , , , These represent the increments in the failure rate per unit length caused by strong winds, lightning, heavy rain, high temperatures, or low temperatures, respectively.

[0108] The above-mentioned increases in failure rates due to strong winds, lightning, rainstorms, high temperatures, or low temperatures are obtained by matching and statistically analyzing historical weather records and historical failure records within the same time window. When the sample size for a certain type of disaster is insufficient, an initial increment coefficient can be given based on the equipment operation procedures, protection action records, and engineering experience thresholds. Then, rolling corrections can be made in S10 using actual failure observation results.

[0109] S2.2 Determine the increment of the strong wind failure rate

[0110] The increment of the failure rate due to strong winds is determined using a piecewise model:

[0111]

[0112] In the formula, For the first Wind speed in the area where the line is located; For the first Critical wind speed for each line; , The parameters are fitted to the failure rate in strong winds.

[0113] S2.3 Determine the increment of failure rate due to lightning, heavy rain, and high or low temperatures.

[0114] The increment of lightning failure rate is determined using a piecewise model:

[0115]

[0116] In the formula, For the first Lightning flash density in the area of ​​the line; The threshold for lightning flash density; These are the fitting parameters for the lightning failure rate.

[0117] For the incremental failure rate due to rainstorms, a threshold-triggered segmented model is used, and the calculation formula is as follows:

[0118]

[0119] in, for Time of the first Rainfall intensity at the route; The threshold for rainfall intensity; This is the rainstorm failure rate intensity coefficient; For the first Correction factor for the equipment withstand level of each line; Weighted by duration of continuous rainfall.

[0120] For the increase in failure rate at high or low temperatures, a temperature limit exceedance model is used: the calculation formula is as follows:

[0121]

[0122] in, for Time of the first The ambient temperature at the route; This refers to the upper limit threshold for high temperature. This is the lower limit threshold for low temperature. This is the temperature-based failure rate intensity coefficient; For the first Correction factor for the equipment withstand level of each line; The duration of temperature exceedance is weighted. The two types of incremental failure rate models mentioned above are used to supplement the impact of extreme weather events other than strong winds and lightning on line failure rates. The model parameters can be continuously calibrated by the online correction module.

[0123] Among them, the disaster trigger threshold, intensity segmentation coefficient, duration weight and equipment tolerance level corresponding to rainstorm, high temperature or low temperature can be determined by historical weather and fault record statistics, equipment operation procedures, protection action records or engineering experience thresholds, and can be rolled over in S10 according to the actual fault observation results.

[0124] S2.4 Calculate the weather exposure factor of the railway line

[0125] Calculate the first Weather exposure factors along the route :

[0126]

[0127] In the formula, The altitude or terrain exposure level of the area where the route is located; The proportion of overhead lines or the degree of exposed lines; This refers to the historical number of line faults. , , , Assigning weights to each exposure factor, And the range of values ​​is It is acceptable As initial weights, when historical failure samples are available, the weights are re-determined through regression fitting, analytic hierarchy process, or expert scoring; the weights of each exposure factor can be determined based on historical failure statistics, operational experience, or expert experience. , , , These are the normalized upper limits corresponding to line length, terrain exposure, overhead ratio or bareness, and number of historical faults, respectively, taking the maximum value or upper limit of the corresponding indicator within the historical sample or engineering design range.

[0128] For example, the line length is normalized using the maximum line length within the assessment area, the terrain exposure level is normalized using the upper limit of the terrain exposure score, the overhead ratio itself is a value between 0 and 1, and the number of historical faults is normalized using the maximum number of faults within the statistical period.

[0129] S2.5 Calculate the corrected failure rate considering differences in line exposure.

[0130] Based on the weather exposure factor of the line , for the The composite failure rate per unit length of each line is corrected to obtain the corrected failure rate:

[0131]

[0132] In the formula, To account for differences in line exposure, the first The line is in The fault rate per unit length at any given time; The larger the value, the higher the exposure of the line under the same extreme weather conditions, and the greater the corresponding corrected failure rate.

[0133] S2.6 Calculate the probability of line faults

[0134] Based on the corrected failure rate Calculate the first The line is within the evaluation time window Internal failure probability:

[0135]

[0136] In the formula, For the first The line is in Time corresponding to evaluation time window Internal failure probability; For the first The length of each line; To evaluate the length of the time window.

[0137] The fault probability is calculated according to the Poisson fault process and is suitable for scenarios with a short evaluation time window and independent line faults. When the evaluation time window is long or there are strongly correlated disasters, the fault probability can be corrected by the scenario correlation correction coefficient.

[0138] S3, Extreme Weather Scene Generation

[0139] This step transforms the continuous weather probability model into a finite set of scenarios that can participate in tidal flow verification and risk calculation. Each scenario includes the values ​​of weather variables and their probabilities of occurrence. Weather variables can be kept consistent through matching within the same historical time period, conditional probability, or correlation sampling, avoiding the independent and arbitrary combination of variables such as wind speed, rainfall, temperature, and lightning.

[0140] First, probabilistic models are established for wind speed, irradiance, lightning, temperature, and rainfall intensity to characterize the statistical characteristics and distribution parameters of each weather variable. Based on this, a historical weather scene sample pool is constructed based on multi-hazard weather records of the same historical period, and extreme weather scenes are generated by sampling or clustering from the historical weather scene sample pool to maintain the correlation between each weather variable.

[0141] S3.1 Wind Speed ​​Probability Model

[0142] Wind speed is modeled using a Weibull distribution, with the following probability density function:

[0143]

[0144] In the formula, For Weibull shape parameters; For Weibull scale parameters.

[0145] S3.2, Solar Irradiance Probability Model

[0146] Solar irradiance is modeled using a Beta distribution. Normalized irradiance is defined as... Its probability density function is:

[0147]

[0148] In the formula, , These are parameters of the Beta distribution; It is the Gamma function; This represents the maximum irradiance.

[0149] S3.3, Probabilistic Model of Lightning Variables

[0150] The lightning variable is modeled jointly using the probability of lightning events and the empirical distribution of lightning flash density. The probability of a lightning event is:

[0151]

[0152] In the formula, This refers to the number of hours of lightning occurrence during the same period in historical statistics. This represents the total number of hours throughout the year.

[0153] When a lightning event occurs, the lightning flash density The lightning flash density is determined based on historical experience distribution of lightning flash density during similar time periods; when no lightning event occurs, the lightning flash density is taken as 0.

[0154] S3.4, Probabilistic Model of Temperature and Rainfall Intensity

[0155] Probabilistic models are established for temperature and rainfall intensity respectively. Temperature can be modeled using a time-segmented normal distribution or an empirical distribution; rainfall intensity can be modeled using a Bernoulli variable indicating whether rainfall occurs plus a Gamma distribution or an empirical distribution of rainfall intensity.

[0156] Among them, the temperature probability model is used to characterize the ambient temperature under different seasons, time periods or weather types. The variation characteristics; the rainfall intensity probability model is used to characterize the probability of rainfall events occurring and the rainfall intensity at the time of occurrence. Distribution characteristics.

[0157] S3.5, Generate extreme weather scenarios

[0158] The probabilistic models for each weather variable established in S3.1 to S3.4 above are used to characterize the statistical characteristics and distribution parameters of each variable, providing a statistical benchmark for subsequent online correction of model parameters. In the scenario generation stage, to maintain the correlation between multiple weather variables and avoid randomly combining each weather variable independently, a scenario sampling and clustering method based on historical weather records of the same period is adopted.

[0159] A historical weather scene sample pool is constructed based on historical meteorological records. Each sample record in the historical weather scene sample pool corresponds to the same historical time period and simultaneously includes wind speed, solar irradiance, lightning density, ambient temperature, and rainfall intensity for that historical time period, denoted as:

[0160]

[0161] In the formula, The first in the historical weather scene sample pool A sample of historical weather scenarios; , , , , These are wind speed, solar irradiance, lightning density, ambient temperature, and rainfall intensity, respectively, for the same historical period.

[0162] Based on the assessment period, season type, weather type, or extreme weather level, historical weather scene samples that match the current assessment conditions are selected from the historical weather scene sample pool. The selected historical weather scene samples are then sampled or clustered to obtain a limited number of extreme weather scenes.

[0163] For the For each extreme weather scenario, the weather scenario vector is:

[0164]

[0165] in, , , , , The first Extreme weather scenarios at any time The wind speed, solar irradiance, lightning density, ambient temperature, and rainfall intensity are all considered.

[0166] because Multiple weather variables in the data come from complete weather records of the same historical period, or from representative scenarios formed based on matching relationships of the same historical period. Therefore, it can maintain the historical correlation between wind speed, irradiance, lightning, temperature and rainfall intensity, and avoid randomly combining the weather variables independently.

[0167] Assign scenario probabilities to extreme weather scenarios obtained through sampling or clustering. The probability of a weather scenario Determined based on the proportion of its corresponding historical sample count to the total number of screened samples:

[0168]

[0169] In the formula, To belong to the Number of historical samples for each weather scenario; Let this be the number of weather scenes generated. The probability of each weather scene satisfies... .

[0170] By combining continuous weather probability models with historical multi-hazard weather records in the above manner, a finite set of weather scenarios that have both probabilistic representation and maintain the correlation of weather variables is generated, which can be used for subsequent distributed energy and load stochastic modeling, joint scenario generation, power flow calculation and risk assessment.

[0171] S4, Distributed Energy Resources and Stochastic Load Modeling

[0172] This step maps the weather scenario obtained in S3 to the operating status of distributed photovoltaic (PV) power, wind power, electric vehicles, energy storage, and loads. PV and wind power are mainly affected by irradiance, temperature, and wind speed; electric vehicles and energy storage mainly reflect adjustability and charge / discharge status; while loads are affected by time of day, temperature, and random demand fluctuations. Through this mapping, weather disturbances no longer only affect the probability of line faults, but also simultaneously affect operating variables such as power sources, loads, and storage.

[0173] S4.1 Photovoltaic Output Model

[0174] The photovoltaic cell temperature, output current, output voltage, fill factor, and photovoltaic array output power are calculated based on solar irradiance and ambient temperature.

[0175] Photovoltaic cell temperature:

[0176]

[0177] Output current:

[0178]

[0179] Output voltage:

[0180]

[0181] Fill factor:

[0182]

[0183] Photovoltaic array output power:

[0184]

[0185] In the formula, This refers to the nominal operating temperature of the photovoltaic cell. This is the short-circuit current; The current temperature coefficient; This is the open-circuit voltage; Voltage temperature coefficient; , The voltage and current at the maximum power point; This refers to the number of photovoltaic cells.

[0186] In the scene ,time Next, , Take them respectively , , obtain the scene Corresponding photovoltaic array output power When photovoltaic models are used for nodes At that time, its rated capacity is taken from the node. Corresponding photovoltaic rated capacity . This represents the equivalent total rated capacity of the photovoltaic modules or photovoltaic arrays connected to the node. The output power calculated from a single photovoltaic module model can be converted into node-level photovoltaic output according to the number of modules connected in series or parallel or the capacity ratio.

[0187] S4.2 Wind Power Output Model

[0188] The wind power output is determined based on the relationship between wind speed and the wind turbine's cut-in wind speed, rated wind speed, and cut-out wind speed.

[0189]

[0190] In the formula, , , These are the cut-in, rated, and cut-out wind velocities, respectively. This refers to the rated power of the fan.

[0191] In the scene ,time Next, the wind speed S3 was obtained , obtain the scene Corresponding wind power output When wind power models are used at nodes At that time, its rated capacity is taken from the node. Corresponding wind power rated capacity . This represents the equivalent total rated capacity of one or more wind turbines connected to the node. The output power calculated by the single-unit wind power model can be converted into node-level wind power output according to the number or capacity ratio of wind turbines.

[0192] S4.3, Electric Vehicle Aggregated Power Model

[0193] Calculate the aggregate power of electric vehicles based on the number of electric vehicles, charging / discharging probability, and charging / discharging power of a single vehicle:

[0194]

[0195] In the formula, For the number of electric vehicles; , For the probability of discharging and charging; , The discharge and charging power of a single unit;

[0196] The probabilities of electric vehicle charging and discharging are determined by historical charging pile records, vehicle arrival / departure statistics, typical travel curves, or dispatch response strategies. Specifically, historical access records are grouped by weekdays / holidays and hourly segments. The number of samples in each time period that a vehicle is in charging, discharging, or disconnected state is counted, and each is divided by the total number of samples in that time period to obtain the charging probability, discharging probability, and disconnection probability. The sum of the probabilities of the three states is 1. If actual measurement records are lacking, typical probability curves can be set according to weekdays, holidays, and peak / off-peak periods, and corrected using actual access data in S10. (Scenario...) ,time Based on the charging and discharging probabilities corresponding to that time period, the scenario is obtained. Corresponding electric vehicle aggregate power When the electric vehicle model is used as a node At that time, its rated charge and discharge capacity is taken from node Corresponding electric vehicle aggregate rated capacity .

[0197] S4.4, Energy Storage and Charge / Discharge Model

[0198] Calculate the current energy of the energy storage system based on its previous energy, charging power, discharging power, and charging / discharging efficiency:

[0199]

[0200] And meet the energy storage constraints:

[0201]

[0202] In the formula, , For energy storage charging and discharging power; , For charging and discharging efficiency; , The minimum and maximum allowable energy for energy storage.

[0203] The energy storage charging and discharging power is determined by the net load following strategy. The node net load is defined as the node load minus the output of photovoltaic power, wind power, and electric vehicle discharge power, plus the electric vehicle charging power. When the node net load is greater than zero and the energy storage energy is above the lower limit, the energy storage discharges, with the discharge power being the smaller of the node net load and the rated discharge power of the energy storage, constrained by the current dischargeable energy of the energy storage. When the node net load is less than zero and the energy storage energy is below the upper limit, the energy storage charges, with the charging power being the smaller of the absolute value of the net load and the rated charging power of the energy storage, constrained by the current rechargeable capacity of the energy storage. When the node net load is zero, or the energy storage energy reaches the upper or lower limit and cannot continue charging or discharging, the energy storage charging and discharging power is zero.

[0204] In the scene ,time Based on the node load, photovoltaic output, wind power output, and aggregated power of electric vehicles corresponding to the scenario, the energy storage charging and discharging power is determined according to the above net load following strategy, and the scenario is obtained. The corresponding energy storage state.

[0205] S4.5, Stochastic Load Model

[0206] Node load is represented using a stochastic model as follows:

[0207]

[0208] In the formula, , For nodes The load mean and standard deviation.

[0209] In the scene Next, time The node loads were obtained by sampling. It is used as the node load input in S5 power flow calculation.

[0210] S5, Weather-Equipment-Load Joint Scenario Generation and Scenario-by-Scenario Power Flow Calculation

[0211] This step combines the aforementioned weather variables, line fault probability, distributed energy output, energy storage / electric vehicle status, and node load into joint operating scenarios. Each joint scenario represents a possible combination of extreme weather, equipment status, and load status, which is used for subsequent AC power flow calculations and risk consequence identification.

[0212] Based on the weather scenario in S3, the distributed energy and load status in S4, and the line fault probability in S2, a weather-equipment-load joint operation scenario is generated. Weather variables, distributed energy output, electric vehicle status, energy storage status, and node load can be generated through Monte Carlo sampling; the line fault probability... It is not directly used to determine the fault state, but rather for probability weighting in subsequent risk calculations. For the consequences of a line fault, in each scenario... and time Below, respectively targeting the first Each line fault event updates the network topology and performs AC power flow calculations to obtain the corresponding comprehensive severity of the consequences.

[0213] Specifically, in the scene and time Set the weather variable as The line fault probability is set as The variables of photovoltaic, wind power, electric vehicles, energy storage and load are respectively taken as the corresponding scenario values, and the AC power flow calculation is performed after updating the network topology for line fault events.

[0214] Before performing AC power flow calculations, the reactive power injection at each node must be determined. The reactive power of the node load is determined based on the node load active power and the preset load power factor. The reactive power corresponding to photovoltaic, wind power, energy storage, and electric vehicles is determined based on their grid-connected inverter power factor, reactive power control mode, or dispatch setpoint. If reactive power regulation is not enabled, distributed energy sources, energy storage, and electric vehicles operate at a unity power factor, and their reactive power output is taken as 0. If reactive power regulation is enabled, their reactive power output is taken within the inverter capacity constraint range and is used as the node reactive power injection. Components of.

[0215] The node power balance constraint is:

[0216]

[0217]

[0218] And verify the state constraints:

[0219]

[0220]

[0221] In the formula, , For nodes Both active and reactive injections; , They are nodes At any moment Active and reactive loads; , They are nodes ,node The voltage amplitude; For nodes With nodes The voltage phase angle difference; , The nodes in the node admittance matrix are respectively With nodes The electrical conductance and susceptance between them; The total number of nodes; , They are nodes The lower limit and upper limit of voltage amplitude; Indicates the first The line at the time Line load or line apparent power; Indicates the first Maximum capacity of each line.

[0222] In scene-by-scene calculations, the above , , , , , All scenes The corresponding scene value.

[0223] For the Line fault event, in scenario and time Next, the first The line is removed from the network topology or set to a fault state, and AC power flow calculation is performed based on the updated network topology to obtain the operational consequences of the line fault event, such as node voltage, line load rate, unsupplied load, and set of islanded or power-out nodes.

[0224] After updating the network topology in response to a line fault event, a connectivity analysis is first performed on the updated network topology to identify the set of isolated or power-depleted nodes that have lost electrical connection with the main power source or available distributed energy resources. For power-depleted nodes, their node loads are included in the unsupplied load. For nodes that are still connected to the power source but cannot be fully supplied due to power flow non-convergence, voltage constraints, or unmet line load constraints, the amount of load to be reduced is determined according to a preset load priority or proportional reduction method, and the reduction amount is included in the unsupplied load. If the network can satisfy the power flow equations, node voltage constraints, and line load constraints after the fault, the unsupplied load of the corresponding node is set to 0.

[0225] This yields the node. In the scene ,time Unsupplied load And further obtain the system's unsupplied load. Number of isolated or power-out nodes and loads in isolated or power-loss areas .

[0226] The output of AC power flow calculation includes: voltage amplitude at each node. Active / reactive power flow and line load rate for each line Unsupplied load Unsupplied load by the system Isolated or power-out node sets, and loads in isolated or power-out areas. The above outputs serve as inputs for the severity of node undervoltage, line overload, load loss, and islanding impact in S6, respectively.

[0227] S6. Construction of Composite Severity Function and Scenario-by-Scenario Severity Calculation

[0228] This step converts different types of operational consequences into comparable severity values ​​based on the power flow calculation results of S5. Node undervoltage severity reflects insufficient voltage safety margin, line overload severity reflects line thermal stability risk, load loss severity reflects power outage losses, and islanding impact severity reflects the damage to the power supply range caused by the fault. These four severity levels together constitute the scenario-level composite severity.

[0229] S6.1, Severity of Node Low Voltage

[0230] The severity of node low voltage is determined based on the relationship between the node voltage magnitude and the reference voltage and the deterministic low voltage limit:

[0231]

[0232] In the formula, Reference voltage; For deterministic low voltage limits; The penalty coefficient for deterministic over-limit regions.

[0233] S6.2, Severity of Line Overload

[0234] Line load rate ;

[0235] The severity of line overload is determined based on the relationship between the line load rate and the safe load rate threshold, and the deterministic overload threshold:

[0236]

[0237] In the formula, The safe load rate threshold; This is a deterministic overload threshold. This is the overload and limit violation penalty coefficient.

[0238] S6.3, Severity of Load Loss

[0239] The severity of load loss is determined based on the ratio between the amount of unsupplied load and the total system load.

[0240]

[0241] In the formula, For nodes In the scene ,time The unsupplied load below; For nodes In the scene ,time The node load below.

[0242] S6.4, Severity of the impact of isolated islands

[0243] The severity of the impact of islanding is determined based on the number of islands or power outage nodes and the load percentage of the islanded or power outage area:

[0244]

[0245] In the formula, This represents the number of isolated or power-out nodes. For isolated or power-loss areas; This represents the total system load. The weighting of the impact of islanded load.

[0246] The above four severity indices are composite consequence evaluation indices constructed by this method. The required parameters are all obtained from the topology, load and line capacity data acquired by S1 and the power flow calculation output results of S5. The initial values ​​of severity weights can be determined based on operational safety standards, load importance levels or historical fault loss data, and are used as initial parameters for subsequent online rolling corrections.

[0247] S6.5, Overall Severity of the Scene

[0248] Scene At any moment The overall severity is as follows:

[0249]

[0250] In the formula, , , , The weights are respectively for the severity of node low voltage, line overload, load loss, and islanding impact. Each weight can be determined by normalizing the safety consequences or economic losses corresponding to low voltage, overload, load loss, and islanding impact, or an initial value can be given by expert scoring or analytic hierarchy process. This represents the total number of lines. The line number in the line overload severity summation item is only used to sum the lines that experienced overload after a fault.

[0251] When the above four severity levels are determined according to the scenario ,time Next When the power flow calculation results corresponding to a line fault event are obtained, the severity of the overall consequences is denoted as: , used to indicate the first The severity of the overall consequences of a line failure event; when the above four severity categories are calculated for the overall operational status of the scenario, the overall severity is recorded as follows: .

[0252] Low voltage limits, line safe load rate thresholds, deterministic overload thresholds, and the weights for each severity level can be determined based on power grid operation safety standards, permissible voltage deviation ranges, line thermal stability constraints, load importance levels, and historical fault loss data.

[0253] S7, System Comprehensive Risk Assessment

[0254] This step multiplies the failure probability by the composite severity for each joint scenario and each assessment period to obtain the scenario risk. Then, it aggregates the risks according to the scenario occurrence probability, assessment period weight, and failure event set to obtain the expected comprehensive risk of the system. This calculation result can be used for risk warning and also as a risk target for subsequent key component ranking and three-objective optimization.

[0255] Single fault event In state The risk value is:

[0256]

[0257] The overall system risk is as follows:

[0258]

[0259] In the formula, The number of faults or disturbances; For state The following fault or disturbance event The probability of occurrence; For fault or disturbance events In state The severity of the corresponding overall consequences.

[0260] For line fault events, the scenario The following are the overall risks:

[0261]

[0262] In the formula, For the first The route in the scene ,time The probability of failure is as follows; For the first Line fault events in the scenario ,time The comprehensive severity of the consequences is obtained after topology update, AC power flow calculation, and composite severity calculation. This represents the total number of lines.

[0263] in, Not determined by the overall severity of the scene It was apportioned afterward, but was targeted at the first... Each line fault event is calculated separately. Specifically, in the scenario... ,time Next, the first The line is set to a fault state and the network topology is updated. AC power flow calculation is performed in step S5. Then, according to the calculation rules for node undervoltage severity, line overload severity, load loss severity, and islanding impact severity in step S6, the result is obtained. The severity of the overall consequences of each line fault event .

[0264] Expected overall risk:

[0265]

[0266] In the formula, Number of scenes; To evaluate the number of time periods. When each scenario is treated with equal probability, The above formula holds true; when different scenarios have different probabilities of occurrence, the expected comprehensive risk can be calculated by using a scenario probability weighting method.

[0267] S8. Identification and reinforcement ranking of risk contributions of key components

[0268] This step is used to identify key components or resources that contribute significantly to system risk. For line components, the risk is directly attributed based on the risks caused by line failure events in various scenarios. For non-line resources such as nodes, distributed energy sources, energy storage, and electric vehicles, a counterfactual marginal risk contribution calculation is used. This involves comparing the expected comprehensive risk of the corresponding object under normal availability with the expected comprehensive risk of the system recalculated after the corresponding object's power supply capacity, access capacity, or availability has decreased.

[0269] S8.1, No. Risk contribution rate of each component

[0270] For line components, the risk contribution value is determined directly based on the expected risk of the line failure event across all scenarios and the assessment period:

[0271]

[0272] In the formula, For the first Expected risks arising from individual circuit components; For the scene The probability of occurrence; Weighting of evaluation periods; For the first The route in the scene ,time The probability of failure is as follows; For the first Line fault events; For the scene ,time The weather scene below; For the first Line fault event In weather scenarios The comprehensive severity of the consequences is obtained through network topology update, S5 AC power flow calculation, and S6 composite severity calculation.

[0273] in, Not determined by the overall severity of the scene It was apportioned afterward, but was targeted at the first... Line fault event The results were obtained by updating the network topology separately and performing power flow calculation and severity calculation.

[0274] For non-line resources such as nodes, distributed energy resources, energy storage, and electric vehicles, the counterfactual marginal risk contribution is used for determination:

[0275]

[0276] In the formula, The expected comprehensive risk of the system under the current solution; To make the first A non-circuit object is placed in a counterfactual downgrade state. Then, re-execute the system expected comprehensive risk obtained from S2-S7. The construction rules are as follows: for capacity-type resources, the rated capacity is reduced by a preset ratio ρ (such as 20% or 50%) or set to zero to form a counterfactual state; for availability-type resources, the availability rate is reduced to form a counterfactual state; ρ is used as a preset parameter or a sensitivity analysis parameter.

[0277] No. The risk contribution rate of each component is:

[0278]

[0279] In the formula, For the first Risk contribution rate of each component or resource; For the first The expected risk arising from a single component or resource; For the first Expected risks arising from components or resources; The total number of components or resources participating in risk ranking.

[0280] S8.2, Key Component Set

[0281] Determine the set of key components based on the risk contribution rate threshold:

[0282]

[0283] In the formula, A collection of key components; The threshold for risk contribution rate.

[0284] S8.3, Unit reinforcement benefit of key components

[0285] For each component in the critical component set, calculate the benefit per unit of reinforcement:

[0286]

[0287] In the formula, For the first Unit reinforcement benefits for key components; The expected comprehensive risk of the system before reinforcement; For the first The expected comprehensive risk of the system is obtained by re-executing S2-S7 after reinforcing key components; For the first The cost of reinforcing key components.

[0288] According to the unit reinforcement revenue The critical components are sorted from high to low to obtain the reinforcement priority of the critical components.

[0289] S9. Synergistic Optimization of Risk-Cost-Resilience Three Objectives

[0290] S9.1 Optimizing the Decision Vector

[0291] Define the optimal decision vector as:

[0292]

[0293] In the formula, , They are nodes Rated capacity of photovoltaic and wind power; Strengthen decision variables for key components.

[0294] When it is necessary to represent the first When making reinforcement decisions for key components, Recorded as .in, For binary variables ,when When it equals 1, it indicates that for the th Reinforcement was implemented on key components; when When the value is 0, it indicates that the first... Reinforcement was implemented on key components.

[0295] After the critical components are reinforced, the corresponding component parameters are adjusted according to the reinforcement type. For line components, reinforcement can reduce their failure rate parameters or increase the line capacity limit. The failure rate parameters or line capacity parameters after reinforcement are determined by the technical parameters of the corresponding reinforcement measures, and then re-enter S2-S7 for failure probability, power flow calculation, severity assessment, and risk calculation.

[0296] S9.2 Total Cost Function

[0297] Construct the total cost function:

[0298]

[0299] The investment cost is as follows:

[0300]

[0301] The operating and maintenance costs are:

[0302]

[0303] The expected cost of power outage losses is:

[0304]

[0305] In the formula, A set of distributed energy types; , For the first Investment cost per unit capacity and operation and maintenance cost per unit of electricity for distributed energy sources; For nodes The Rated capacity of distributed energy resources; For the scene ,time Next node The Distributed energy output; For the scene The probability of; Cost per unit of electricity not supplied; The system unsupplied load output by S5 power flow calculation; For the first The cost of reinforcing key components; For the first Decision variables for strengthening key components; The total number of components involved in reinforcement or risk prioritization.

[0306] S9.3, System Resilience Indicators

[0307] Constructing system resilience indicators:

[0308]

[0309] In the formula, For the scene The system at time Performance level; For the system in a undisturbed state at time 10:00 Performance level; , The start and end times of the impact of extreme weather.

[0310] System performance level Determined based on load supply ratio:

[0311]

[0312] In the formula, For the scene ,time The system is not supplying load; For the scene ,time The total system load is as follows. A larger value indicates better system power supply performance. System performance level under undisturbed conditions. It can be determined based on the load supply ratio under undisturbed conditions; when all loads can be powered under undisturbed conditions, The value is 1.

[0313] S9.4, Three-Objective Optimization Model

[0314] Establish a three-objective optimization model:

[0315]

[0316] The constraints are:

[0317]

[0318]

[0319]

[0320] In the formula, For nodes The Rated capacity of distributed energy resources; For nodes The Maximum capacity of distributed energy resources; For the first The cost of reinforcing individual components; For the first The hardening decision variables for each component; To reinforce the budget ceiling.

[0321] The algorithm employs a multi-objective genetic algorithm or other intelligent optimization algorithm for iterative search. Each candidate solution must undergo S2-S7 processes, including scenario generation, failure rate correction, power flow calculation, severity assessment, and risk calculation, to output a Pareto frontier solution. The recommended solution is then selected based on risk preference, investment budget, and resilience requirements.

[0322] S10, Online rolling correction of model parameters

[0323] The failure rate parameter is corrected online using actual weather, fault, and operational data.

[0324]

[0325] Loss function:

[0326]

[0327] In the formula, This is a parameter for the failure rate in strong winds; To learn step size; For the first This route at a historical moment Observations indicating whether a fault has occurred; The failure probability calculated for the model; This represents the length of the historical samples. Similarly, the severity weights and scenario probabilities are adjusted using a rolling adjustment mechanism.

[0328] Online rolling corrections are used not only for strong wind failure rate parameters, but also for failure rate parameters, severity weights, and scenario probabilities for lightning, heavy rain, high and low temperatures. After each extreme weather event, the actual failures, actual power flow exceedances, actual unsupplied loads, and scheduling execution effects are compared with the model calculation results to update the parameters, enabling the model to adapt to different regions, seasons, and grid conditions.

[0329] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A collaborative optimization method for power grid risk resilience driven by multi-hazard exposure, characterized in that, include: Acquire extreme weather data, power grid topology and line parameters, distributed energy and load data, electric vehicle and energy storage data, and historical fault records to construct an operating environment state vector; Based on the wind speed, lightning density, rainfall intensity, and ambient temperature data from extreme weather, the failure rate increment is determined, the line exposure factor is determined, the composite failure rate per unit length of the line is corrected, and the line failure probability is obtained. Establish probabilistic models for various weather variables, generate extreme weather scenarios, determine scenario-based operational states based on extreme weather scenarios, and construct joint operational scenarios; In each joint operation scenario, the power grid topology is updated and AC power flow calculations are performed for line fault events. Based on the power flow calculation results, the severity of node low voltage, line overload, load loss and islanding impact are determined. The severity of the comprehensive consequences of line fault events is calculated. The line fault probability is combined with the corresponding severity of comprehensive consequences to calculate the expected comprehensive risk of the system. Perform risk contribution decomposition, identify key components with risk contribution rates exceeding the threshold, calculate unit reinforcement benefits and determine reinforcement priorities; use key components and their reinforcement priorities as inputs for key component reinforcement decisions; With the expected comprehensive risk, total cost, and system resilience as optimization objectives, a three-objective collaborative optimization model is established, which includes distributed energy capacity allocation and key component reinforcement decisions. The three-objective collaborative optimization model is solved to obtain the output of the power grid collaborative optimization scheme.

2. The multi-hazard exposure-driven power grid risk resilience collaborative optimization method according to claim 1, characterized in that, The failure rate increment includes the failure rate increments for strong winds, lightning, heavy rain, and high or low temperature failures triggered by ambient temperature. Among them, the failure rate increment for strong winds is zero when the wind speed does not exceed the critical wind speed, and increases with the magnitude of the wind speed exceeding the limit after the wind speed exceeds the critical wind speed; the failure rate increment for lightning is zero when the lightning ground flash density does not exceed the threshold, and increases with the magnitude of the lightning ground flash density exceeding the limit after the lightning ground flash density exceeds the threshold; the failure rate increments for heavy rain, high or low temperature are determined based on the disaster trigger threshold, intensity segmentation coefficient, duration, and equipment tolerance level.

3. The multi-hazard exposure-driven power grid risk resilience collaborative optimization method according to claim 1, characterized in that, The line exposure factor is determined by weighting the normalized results of line length, terrain exposure level, overhead ratio, and number of historical faults, with the sum of the weights of each exposure factor being 1.

4. The method for collaborative optimization of power grid risk resilience driven by multi-hazard exposure according to claim 1, characterized in that, The generation of extreme weather scenarios specifically includes: establishing probability models for wind speed, solar irradiance, lightning variables, ambient temperature, and rainfall intensity, and performing scenario sampling or clustering based on historical weather records for the same period to obtain a set of extreme weather scenarios including wind speed, solar irradiance, lightning ground flash density, ambient temperature, and rainfall intensity, and determining the probability of occurrence of each extreme weather scenario.

5. The multi-hazard exposure-driven power grid risk resilience collaborative optimization method according to claim 1, characterized in that, Specific methods for determining scenario-based operational status in conjunction with extreme weather scenarios include: Based on the solar irradiance and ambient temperature of the corresponding scenario, the photovoltaic output is calculated through photovoltaic cell temperature, output current, voltage, and fill factor; based on the wind speed of the corresponding scenario, the wind power output is determined according to the segmented relationship of cut-in wind speed, rated wind speed, and cut-out wind speed; the aggregate power of electric vehicles is calculated based on the number of electric vehicles connected, the charging and discharging probability during the time period, and the charging and discharging power of a single vehicle; a net load following strategy is adopted, and the energy storage charging and discharging power and time-series energy status are determined by combining the node net load status and the upper and lower limits of energy storage; the node load is sampled using a normal distribution random model to obtain the scenario load value, thereby realizing the synchronous mapping of weather disturbances to the source, load, and storage operation variables.

6. The method for collaborative optimization of power grid risk resilience driven by multi-hazard exposure according to claim 1, characterized in that, The expected comprehensive risk of the calculation system includes: in each joint operation scenario and each evaluation period, for each line fault event in the set of line fault events, setting the corresponding line to a fault state and updating the power grid topology, performing AC power flow calculation, and obtaining the comprehensive consequence severity of the line fault event based on the severity of node low voltage, line overload, load loss, and islanding impact; combining the line fault probability of the line fault event with the corresponding comprehensive consequence severity to obtain the risk value of the line fault event; summarizing the risk values ​​of each line fault event in the set of line fault events to obtain the scenario comprehensive risk under the corresponding joint operation scenario and evaluation period; and finally summarizing the scenario comprehensive risks under each joint operation scenario and each evaluation period to obtain the expected comprehensive risk of the system.

7. The method for collaborative optimization of power grid risk resilience driven by multi-hazard exposure according to claim 1, characterized in that, The process of identifying key components with risk contribution rates exceeding a threshold, calculating unit reinforcement benefits, and determining reinforcement priorities includes: for line components, determining their risk contribution value based on the expected risk generated by the corresponding line fault event across all joint operation scenarios and evaluation periods; for nodes, distributed energy resources, energy storage, and electric vehicle resources, using a counterfactual marginal risk contribution method, comparing the expected comprehensive risk of the corresponding resource under normal availability with the expected comprehensive risk of the corresponding resource under reduced capacity, out of operation, or reduced availability, and determining its risk contribution value based on the difference; determining the corresponding risk contribution rate based on the proportion of each component's or resource's risk contribution value to the total risk contribution value of all components or resources participating in risk ranking; screening components or resources with risk contribution rates exceeding a preset threshold as key components; for each key component, determining the unit reinforcement benefit based on the reduction between the expected comprehensive risk of the system before reinforcement and the expected comprehensive risk of the system after reinforcement, as well as the reinforcement cost of the key component, and determining the reinforcement priority of key components according to the unit reinforcement benefit from high to low.

8. The method for collaborative optimization of power grid risk resilience driven by multi-hazard exposure according to claim 1, characterized in that, In the three-objective collaborative optimization model, the total cost consists of three parts: the investment cost of distributed energy capacity configuration and key component reinforcement, the operation and maintenance cost of distributed energy, and the expected power outage loss cost. The system resilience index is represented by the ratio of the integral of the actual load supply ratio of the system during the period of extreme weather impact to the integral of the load supply ratio under undisturbed conditions.

9. The method for collaborative optimization of power grid risk resilience driven by multi-hazard exposure according to claim 1, characterized in that, It also includes online rolling correction, specifically: after each extreme weather event, the actual line fault observation results, operation status data and model calculation results are compared. The mean square error between the fault observation values ​​and the fault probability calculated by the model within the historical period is used as the loss function. The fault rate parameters corresponding to strong wind, lightning, rainstorm, high temperature and low temperature are iteratively updated using gradient descent. At the same time, the severity weight and weather scenario probability are corrected simultaneously.

10. A multi-hazard exposure-driven power grid risk resilience collaborative optimization system, characterized in that, include: The operating environment perception module is used to acquire extreme weather data, power grid topology and line parameters, distributed energy and load data, electric vehicle and energy storage data, and historical fault records to construct an operating environment state vector. The line fault probability calculation module is used to determine the fault rate increment based on wind speed, lightning density, rainfall intensity, and ambient temperature in extreme weather data, determine the line exposure factor, correct the composite fault rate per unit length of the line, and obtain the line fault probability. The joint operation scenario generation module is used to establish probability models for various weather variables, generate extreme weather scenarios, determine the scenario-based operation status based on the extreme weather scenarios, and construct joint operation scenarios. The fault scenario power flow calculation module is used to update the power grid topology and perform AC power flow calculations for line fault events in various joint operation scenarios. The system risk quantification module is used to determine the severity of node low voltage, line overload, load loss and islanding impact based on power flow calculation results, calculate the comprehensive consequence severity corresponding to line fault events, and combine the line fault probability with the corresponding comprehensive consequence severity to calculate the expected comprehensive risk of the system. The critical component identification module is used to decompose risk contribution, identify critical components with a risk contribution rate higher than the threshold, calculate the unit reinforcement benefit and determine the reinforcement priority, and use the critical components and their reinforcement priorities as inputs for critical component reinforcement decisions. The collaborative optimization model construction module is used to establish a three-objective collaborative optimization model with the system's expected comprehensive risk, total cost, and system resilience as optimization objectives, including distributed energy capacity configuration and key component reinforcement decisions. The optimization solution and output module is used to solve the three-objective collaborative optimization model and obtain the output of the power grid collaborative optimization scheme.

Images