Power distribution network reliability dynamic hierarchical early warning system and method of multi-source disturbance

Abstract

The application discloses a power distribution network reliability dynamic hierarchical early warning system and method of multi-source disturbance, and belongs to the field of power system reliability evaluation. The early warning system comprises a multi-source perception layer, a dynamic evaluation layer, a hierarchical early warning layer and a resource control layer which are sequentially and communicatively connected. The multi-source perception layer is used for integrating multi-element real-time data to provide comprehensive disturbance information for the dynamic evaluation layer. The dynamic evaluation layer constructs a comprehensive disturbance index through a disturbance factor fusion engine based on the output of the multi-source perception layer. The hierarchical early warning layer automatically matches an early warning level and a control strategy according to the evaluation result of the dynamic evaluation layer. The resource control layer serves as an instruction execution terminal. Through the cooperative construction of a dynamic disturbance factor library, multi-dimensional real-time data such as weather, DER output and load clustering are fused, multi-source coupling risks which are difficult to capture in traditional static analysis are effectively quantified, and the fault prediction accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of power system reliability assessment, and particularly to the scenario of large-scale distributed energy access. It proposes a dynamic hierarchical early warning system and method for distribution network reliability that integrates the fluctuation characteristics of distributed resource clusters, environmental disturbances, and load change characteristics. Background Technology

[0002] With the acceleration of energy transition, large-scale grid connection of distributed resources has become an inevitable trend in the development of distribution networks. According to the State Grid's 2024 New Energy Consumption Report, in areas where the penetration rate of distributed photovoltaic exceeds 30%, the failure rate of distribution networks increases by 42% year-on-year, of which multi-source disturbance coupling failures account for as high as 67%, highlighting the shortcomings of traditional early warning systems in terms of response lag. However, existing reliability assessment technologies are ill-suited to the new challenges brought about by the high proportion of distributed energy access. Traditional methods mainly rely on historical outage data for static statistical analysis, which is insufficient to quantify the real-time risks caused by the output fluctuations of intermittent resources such as distributed photovoltaic and wind power. Especially in scenarios with multiple disturbances coupled by sudden weather changes and rapid load changes, existing models cannot accurately characterize the power coordination fluctuation characteristics caused by the cluster effect of distributed resources. At the same time, current distribution network reliability improvement strategies are mostly based on post-event assessment or offline simulation, with response speed lagging behind the fault evolution process. This makes it difficult to trigger emergency control measures such as load shedding and energy storage scheduling in a timely manner, resulting in decision-making blind spots when the system faces cascading faults such as voltage over-limit and protection malfunction. In addition, the complexity of distribution network operation increases dramatically after large-scale distributed resource grid connection, and the existing assessment system has not yet established a dynamic mapping mechanism between multi-dimensional disturbance factors such as environmental factors, grid structure, and resource fluctuations and reliability indicators, which restricts the accuracy of pre-event warnings and makes the resilience improvement of distribution networks in the face of high-penetration distributed energy access face technical bottlenecks. Summary of the Invention

[0003] This invention aims to solve the problems of response lag, lack of cluster fluctuation quantification, and mismatch between control strategy and fault evolution speed in traditional distribution network reliability early warning systems and methods. Therefore, it provides a dynamic hierarchical early warning system and method for distribution network reliability with multi-source disturbances.

[0004] A dynamic hierarchical early warning system for the reliability of a distribution network with multi-source disturbances includes a multi-source sensing layer, a dynamic assessment layer, a hierarchical early warning layer, and a resource control layer that are connected in sequence via communication. The multi-source sensing layer is used to integrate multi-source real-time data, including: wind speed obtained through the meteorological bureau API. Real-time light intensity Calculate the ambient temperature T and ambient humidity H, and calculate the meteorological disturbance index. It is used to measure the severity of the impact of different meteorological factors on system disturbances; in, For maximum light intensity, H represents the reference temperature; H represents the ambient humidity. For reference humidity, These are the weighting coefficients for each item; Real-time photovoltaic power output obtained through the power grid data acquisition and monitoring system SC Real-time power output of the blower Based on this, the fluctuation entropy of the DER cluster is calculated. Where N is the number of DER distributed energy nodes; This refers to the power variation of a single DER unit. This represents the total real-time power of the entire DER cluster. And load change rate identification based on Advanced Measurement Architecture (AMI) It also completes the clustering of abnormal user groups, providing comprehensive disturbance information for the dynamic evaluation layer; The dynamic evaluation layer relies on the output of the multi-source sensing layer and constructs a comprehensive disturbance index through a disturbance factor fusion engine. In the formula, The vulnerability coefficient of the space frame structure; , , These are adaptive weighting coefficients that are dynamically adjusted using the tanh function. The rate of change of wave entropy; The value ranges from [0,1], with larger values ​​indicating higher risks. To accelerate load changes, and simultaneously utilize a chain-like risk transmission model, a system is established to address power fluctuations. ΔP To voltage deviation ΔV From protecting the probability of false activation Pm Ultimately, the fault evolution path predicted by SAIDI and SAIFI will be realized, where SAIDI represents the system average outage duration and SAIFI represents the system average outage frequency index, providing a quantitative basis for the graded early warning layer. The hierarchical early warning layer automatically matches the early warning level with the control strategy based on the evaluation results of the dynamic evaluation layer; The resource control layer, as the instruction execution terminal, is equipped with a fast instruction generator based on a field-programmable gate array (FPGA), which can generate GOOSE control messages within 200ms and send them to the field station equipment. It is also equipped with a policy verification module, which verifies the control policy through real-time digital simulation (RTDS) to ensure the reliability of early warning actions and the effectiveness of system defense.

[0005] Furthermore, the correspondence between the warning levels and control strategies is as follows: when A red alert is triggered if the predicted average outage duration (SAIDI) exceeds 3 hours, or if the SAIDI is 0.8 or higher, and load shedding and energy storage discharge are executed. When 0.5 < If the value is ≤0.8 or SAIFI exceeds the limit, a yellow warning is triggered, and the microgrid is switched to islanded operation. Regarding voltage deviation Δ V In cases where the probability exceeds 10%, a blue alert is triggered, and the Static Var Generator (SVG) is activated for reactive power compensation, thereby achieving pre-emptive graded blocking.

[0006] A dynamic hierarchical early warning method for distribution network reliability under multi-source disturbances is proposed, which employs a dynamic hierarchical early warning system for distribution network reliability under multi-source disturbances and includes the following steps: Step 1, Multi-source perturbation feature extraction: First, the meteorological disturbance index is extracted and calculated in real time through a multi-source sensing layer. DER cluster fluctuation entropy and load mutation rate based on Advanced Measurement Architecture (AMI) measurement identification. This provides quantitative input for risk assessment, including the meteorological disturbance index. The calculation formula is as follows: In the formula, For wind speed, Real-time light intensity; is the maximum light intensity; T is the ambient temperature; is the reference temperature; H is the ambient humidity; is the reference humidity; are the weighting coefficients for each item, used to measure the severity of the impact of different meteorological factors on system disturbances; DER Cluster Fluctuation Entropy The calculation formula is as follows: ; In the formula, N represents the number of DER distributed energy nodes; This refers to the power variation of a single DER unit. This represents the total real-time power of the entire DER cluster. Step 2, Calculation of dynamic risk entropy: The dynamic evaluation layer acquires data from the multi-source perception layer and integrates the vulnerability coefficient of the grid structure. Construct a comprehensive disturbance index: In the formula, , , These are adaptive weighting coefficients that are dynamically adjusted using the tanh function. The rate of change of wave entropy This refers to the acceleration due to load change. It is used to comprehensively reflect the real-time risk level of a system under the combined effects of multiple disturbances. The larger the structure, the more sensitive it is to DER power fluctuations. When the value is >0.8, a 10% change in DER output will cause a voltage deviation of more than 5%; The value ranges from [0,1], with larger values ​​indicating higher risks. Step 3, Tiered Strategy Matching: Based on the chain transmission model, establish a system from power fluctuations ΔP To voltage deviation ΔV From protecting the probability of false activation Pm Predict the system's average outage duration (SAIDI) and average outage frequency (SAIFI); The tiered strategy matching module dynamically triggers control strategies based on risk entropy values ​​and reliability index prediction results. like If the value is >0.8 or the predicted average outage duration SAIDI exceeds 3 hours, a red alert will be activated, and load shedding and energy storage discharge will be performed. If 0.5 < If the value is ≤0.8 or SAIFI exceeds the limit, a yellow alert will be activated, and the microgrid will be switched to islanded operation. In other cases, a blue alert is activated, and SVG reactive power compensation is implemented to achieve multi-level coordinated control that matches the speed of fault evolution. Step 4, Resource Control The resource control layer aims to achieve millisecond-level control command execution and dynamic policy verification to ensure real-time and reliable defense effects. The resource control layer uses a field-programmable gate array (FPGA) as the core control unit. When the hierarchical policy matching module outputs the warning level, the FPGA generates the corresponding control command according to the preset "warning level-command" mapping table, and encapsulates the command code, target device address and priority into a GOOSE message that conforms to the IEC 61850 standard. The instructions are sent to each execution terminal, including the distribution transformer terminal (DTU), energy storage controller, and SVG controller, via the fiber optic ring network with a transmission delay of less than 10ms. The execution terminals parse the messages in real time and trigger the operation. The resource control layer also has a strategy verification and feedback mechanism. By introducing the distribution network PMU to collect system status data in real time, the actual voltage, power and other data are compared with the expected status. If the deviation exceeds the threshold, the strategy parameters are dynamically adjusted and the instructions are regenerated. At the same time, the data of the whole process is recorded in the database for strategy backtracking and optimization, forming a closed-loop control system of "execution-monitoring-verification-adjustment", which greatly improves the control accuracy and robustness of the system in real disturbance environment.

[0007] Beneficial effects: I. By collaboratively constructing a dynamic disturbance factor library and integrating multi-dimensional real-time data such as meteorological data, DER output data, and load clustering data, we can effectively quantify the multi-source coupling risks that are difficult to capture by traditional static analysis and improve the accuracy of fault prediction. II. Mechanistic analysis based on the chain transmission model accurately correlates the causal relationship between power fluctuations, voltage deviations and protection maloperations, breaking through the attribution bottleneck of "distributed resources inducing reliability events" and providing theoretical support for graded early warning. Third, through a three-level strategy dynamic matching mechanism, accurate response is achieved in scenarios such as multi-source disturbances, filling the application gap of pre-assessment criteria in the field of real-time control, and simultaneously optimizing the early warning speed and strategy effectiveness. Attached Figure Description

[0008] Figure 1 This is a system architecture diagram of a dynamic hierarchical early warning system for the reliability of a distribution network subject to multi-source disturbances.

[0009] Figure 2 This is a module interaction diagram of a dynamic hierarchical early warning system for the reliability of a distribution network subject to multi-source disturbances.

[0010] Figure 3 It is a chain-like risk transmission model for a dynamic hierarchical early warning system for the reliability of distribution networks with multi-source disturbances. Detailed Implementation

[0011] A dynamic hierarchical early warning system for the reliability of a distribution network with multi-source disturbances includes a multi-source sensing layer, a dynamic assessment layer, a hierarchical early warning layer, and a resource control layer that are connected in sequence via communication. The multi-source sensing layer is used to integrate multi-source real-time data, including: wind speed obtained through the meteorological bureau API. Real-time light intensity Calculate the ambient temperature T and ambient humidity H, and calculate the meteorological disturbance index. It is used to measure the severity of the impact of different meteorological factors on system disturbances; in, For maximum light intensity, H represents the reference temperature; H represents the ambient humidity. For reference humidity, These are the weighting coefficients for each item; Real-time photovoltaic power output obtained through the power grid data acquisition and monitoring system SC Real-time power output of the blower Based on this, the fluctuation entropy of the DER cluster is calculated. Where N is the number of DER distributed energy nodes; This refers to the power variation of a single DER unit. This represents the total real-time power of the entire DER cluster. And load change rate identification based on Advanced Measurement Architecture (AMI) It also completes the clustering of abnormal user groups, providing comprehensive disturbance information for the dynamic evaluation layer; The dynamic evaluation layer relies on the output of the multi-source sensing layer and constructs a comprehensive disturbance index through a disturbance factor fusion engine. In the formula, The vulnerability coefficient of the space frame structure; , , These are adaptive weighting coefficients that are dynamically adjusted using the tanh function. The rate of change of wave entropy; The value ranges from [0,1], with larger values ​​indicating higher risks. To accelerate load changes, and simultaneously utilize a chain-like risk transmission model, a system is established to address power fluctuations. ΔP To voltage deviation ΔV From protecting the probability of false activation Pm Ultimately, the fault evolution path predicted by SAIDI and SAIFI will be realized, where SAIDI represents the system average outage duration and SAIFI represents the system average outage frequency index, providing a quantitative basis for the graded early warning layer. The hierarchical early warning layer automatically matches the early warning level with the control strategy based on the evaluation results of the dynamic evaluation layer; The hierarchical early warning layer automatically matches early warning levels with control strategies based on the evaluation results of the dynamic evaluation layer. The correspondence between early warning levels and control strategies is as follows: when A red alert is triggered if the average outage duration (SAIDI) is >0.8 or exceeds 3 hours, and load shedding and energy storage discharge are executed. When 0.5 < If the value is ≤0.8 or SAIFI exceeds the limit, a yellow warning is triggered, and the microgrid is switched to islanded operation. Regarding voltage deviation Δ V In cases where the probability exceeds 10%, a blue alert is triggered, and the Static Var Generator (SVG) is activated for reactive power compensation, thereby achieving pre-emptive graded blocking.

[0012] The resource control layer, as the instruction execution terminal, is equipped with a fast instruction generator based on a field-programmable gate array (FPGA), which can generate GOOSE control messages within 200ms and send them to the field station equipment. It is also equipped with a policy verification module, which verifies the control policy through real-time digital simulation (RTDS) to ensure the reliability of early warning actions and the effectiveness of system defense.

[0013] A dynamic hierarchical early warning method for distribution network reliability under multi-source disturbances is proposed, which employs a dynamic hierarchical early warning system for distribution network reliability under multi-source disturbances and includes the following steps: Step 1, Multi-source perturbation feature extraction: First, the meteorological disturbance index is extracted and calculated in real time through a multi-source sensing layer. DER cluster fluctuation entropy and load mutation rate based on Advanced Measurement Architecture (AMI) measurement identification. This provides quantitative input for risk assessment, including the meteorological disturbance index. The calculation formula is as follows: In the formula, For wind speed, Real-time light intensity; Maximum light intensity; T is ambient temperature; is reference temperature; H is ambient humidity; is reference humidity; are the weighting coefficients for each item. The stronger the real-time light intensity... The closer to The closer this value is to 0, the more beneficial it is to photovoltaic output and the lower the degree of degradation; real-time light intensity The weaker, that is The value approaches 0; the closer this value is to 1, the greater the decline in photovoltaic output and the more severe the deterioration. It is used to measure the severity of the impact of different meteorological factors on system disturbances. DER Cluster Fluctuation Entropy The calculation formula is as follows: ; In the formula, N represents the number of DER distributed energy nodes; This refers to the power variation of a single DER unit. This represents the total real-time power of the entire DER cluster. Step 2, Calculation of dynamic risk entropy: The dynamic evaluation layer acquires data from the multi-source perception layer and integrates the vulnerability coefficient of the grid structure. Construct a comprehensive disturbance index: In the formula, , , These are adaptive weighting coefficients that are dynamically adjusted using the tanh function. The rate of change of wave entropy This refers to the acceleration due to load change. It is used to comprehensively reflect the real-time risk level of a system under the combined effects of multiple disturbances. The larger the structure, the more sensitive it is to DER power fluctuations. When the value is >0.8, a 10% change in DER output will cause a voltage deviation of more than 5%; The value ranges from [0,1], with larger values ​​indicating higher risks. Step 3, Tiered Strategy Matching: Based on the chain transmission model, establish a system from power fluctuations ΔP To voltage deviation ΔV From protecting the probability of false activation Pm Predict the system's average outage duration (SAIDI) and average outage frequency (SAIFI); The tiered strategy matching module dynamically triggers control strategies based on risk entropy values ​​and reliability index prediction results. like If the value is >0.8 or the predicted average outage duration SAIDI exceeds 3 hours, a red alert will be activated, and load shedding and energy storage discharge will be performed. It should be noted that energy storage discharge can only be performed when the energy storage state of charge (SOC) is >80%; otherwise, only load shedding will be performed. If 0.5 < If the value is ≤0.8 or SAIFI exceeds the limit, a yellow alert will be activated, and the microgrid will be switched to islanded operation. In other cases, a blue alert is activated, and SVG reactive power compensation is implemented to achieve multi-level coordinated control that matches the speed of fault evolution. Step 4, Resource Control The resource control layer aims to achieve millisecond-level control command execution and dynamic policy verification to ensure real-time and reliable defense effects. The resource control layer uses a field-programmable gate array (FPGA) as the core control unit. When the hierarchical policy matching module outputs the warning level, the FPGA generates the corresponding control command according to the preset "warning level-command" mapping table, and encapsulates the command code, target device address and priority into a GOOSE message that conforms to the IEC 61850 standard. The instructions are sent to each execution terminal, including the distribution transformer terminal (DTU), energy storage controller, and SVG controller, via the fiber optic ring network with a transmission delay of less than 10ms. The execution terminals parse the messages in real time and trigger the operation. The resource control layer also has a strategy verification and feedback mechanism. By introducing the distribution network PMU to collect system status data in real time, the actual voltage, power and other data are compared with the expected status. If the deviation exceeds the threshold, the strategy parameters are dynamically adjusted and the instructions are regenerated. At the same time, the data of the whole process is recorded in the database for strategy backtracking and optimization, forming a closed-loop control system of "execution-monitoring-verification-adjustment", which greatly improves the control accuracy and robustness of the system in real disturbance environment.

[0014] Taking a typhoon as an example, the application scenario is a power distribution network in a development zone with a photovoltaic penetration rate of 38%. The specific implementation process is as follows: At the multi-source sensing layer, the meteorological module detected in real time a sudden increase in wind speed to 25 m / s and a sharp drop in light intensity to 50 W / m², while temperature and humidity also showed significant abnormal fluctuations deviating from reference values. This data was synchronously input into the weather deterioration index. The fusion calculations performed in the model caused its values ​​to rise sharply. Simultaneously, the DER module detected drastic fluctuations in photovoltaic output and calculated the cluster fluctuation entropy accordingly. The value jumped sharply from 0.2 to 0.9. The multi-source sensing layer thus accurately captured the complex strong disturbance characteristics caused by extreme weather and uncontrolled power output from new energy sources.

[0015] At the dynamic assessment layer, the disturbance factor fusion engine is immediately activated. Instead of a simple summation, it dynamically weighs the impacts of drastic fluctuations in entropy, the load change acceleration identified by AMI measurements, and the weather deterioration index using adaptive weighting coefficients, and then multiplies the result by the grid structure vulnerability coefficient. Finally, the system calculates the comprehensive disturbance index. The value has exceeded the red alert threshold of 0.8, and the chain risk transmission model predicts that the average outage duration (SAIDI) will reach 4.2 hours. Based on this, the system determines that the distribution network is facing a high-risk operating state. Initial power disturbances are highly likely to trigger local voltage exceedances and may further cause cascading malfunctions in protection devices, ultimately escalating into a power outage. The tiered early warning layer triggers a red alert, and the resource control layer responds immediately: the FPGA control chip generates and issues a GOOSE command within 200 milliseconds, executing coordinated control measures to cut off 30MW of non-critical loads and discharge the energy storage system at full power. Real-time digital simulation using RTDS verifies that after control, the system voltage quickly recovers to 0.95 pu, and the probability of protection malfunctions decreases to below 5%, effectively suppressing the risk of cascading failures. This demonstrates the effectiveness and reliability of the invention in dealing with extreme disaster scenarios in high-penetration new energy distribution networks.

[0016] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any equivalent substitutions and modifications made by those skilled in the art without departing from the guidance of the present invention should be considered to fall within the protection scope of the present invention.

Claims

1. A dynamic hierarchical early warning system for the reliability of a distribution network under multi-source disturbances, characterized in that, It includes a multi-source sensing layer, a dynamic evaluation layer, a hierarchical early warning layer, and a resource control layer with sequential communication connections; The multi-source sensing layer is used to integrate multi-source real-time data, including: wind speed obtained through the meteorological bureau API. Real-time light intensity Calculate the ambient temperature T and ambient humidity H, and calculate the meteorological disturbance index. It is used to measure the severity of the impact of different meteorological factors on system disturbances; in, For maximum light intensity, H represents the reference temperature; H represents the ambient humidity. For reference humidity, These are the weighting coefficients for each item; Real-time photovoltaic power output obtained through the power grid data acquisition and monitoring system SC Real-time power output of the blower Based on this, the fluctuation entropy of the DER cluster is calculated. Where N is the number of DER distributed energy nodes; This refers to the power variation of a single DER unit. This represents the total real-time power of the entire DER cluster. And load change rate identification based on Advanced Measurement Architecture (AMI) It also completes the clustering of abnormal user groups, providing comprehensive disturbance information for the dynamic evaluation layer; The dynamic evaluation layer relies on the output of the multi-source sensing layer and constructs a comprehensive disturbance index through a disturbance factor fusion engine. In the formula, The vulnerability coefficient of the space frame structure; , , These are adaptive weighting coefficients that are dynamically adjusted using the tanh function. The rate of change of wave entropy; The value ranges from [0,1], with larger values ​​indicating higher risks. To accelerate load changes, and simultaneously utilize a chain-like risk transmission model, a system is established to address power fluctuations. ΔP To voltage deviation ΔV From protecting the probability of false activation Pm Ultimately, the fault evolution path predicted by SAIDI and SAIFI will be realized, where SAIDI represents the system average outage duration and SAIFI represents the system average outage frequency index, providing a quantitative basis for the graded early warning layer. The hierarchical early warning layer automatically matches the early warning level with the control strategy based on the evaluation results of the dynamic evaluation layer; The resource control layer, as the instruction execution terminal, is equipped with a fast instruction generator based on a field-programmable gate array (FPGA), which can generate GOOSE control messages within 200ms and send them to the field station equipment. It is also equipped with a policy verification module, which verifies the control policy through real-time digital simulation (RTDS) to ensure the reliability of early warning actions and the effectiveness of system defense.

2. The dynamic hierarchical early warning system for distribution network reliability based on multi-source disturbances according to claim 1, characterized in that, The correspondence between the warning levels and control strategies is as follows: when A red alert is triggered if the predicted average outage duration (SAIDI) exceeds 3 hours, or if the SAIDI is 0.8 or higher, and load shedding and energy storage discharge are executed. When 0.5 < If the value is ≤0.8 or SAIFI exceeds the limit, a yellow warning is triggered, and the microgrid is switched to islanded operation. Regarding voltage deviation Δ V In cases where the probability exceeds 10%, a blue alert is triggered, and the Static Var Generator (SVG) is activated for reactive power compensation, thereby achieving pre-emptive graded blocking.

3. A dynamic hierarchical early warning method for distribution network reliability under multi-source disturbances, employing the dynamic hierarchical early warning system for distribution network reliability as described in claim 2, characterized in that... Includes the following steps: Step 1, Multi-source perturbation feature extraction: First, the meteorological disturbance index is extracted and calculated in real time through a multi-source sensing layer. DER cluster fluctuation entropy and load mutation rate based on Advanced Measurement Architecture (AMI) measurement identification. This provides quantitative input for risk assessment, including the meteorological disturbance index. The calculation formula is as follows: In the formula, For wind speed, Real-time light intensity; is the maximum light intensity; T is the ambient temperature; is the reference temperature; H is the ambient humidity; is the reference humidity; are the weighting coefficients for each item, used to measure the severity of the impact of different meteorological factors on system disturbances; DER Cluster Fluctuation Entropy The calculation formula is as follows: ； In the formula, N represents the number of DER distributed energy nodes; This refers to the power variation of a single DER unit. This represents the total real-time power of the entire DER cluster. Step 2, Dynamic Risk Entropy Calculation: The dynamic evaluation layer acquires data from the multi-source perception layer and integrates the vulnerability coefficient of the grid structure. Construct a comprehensive disturbance index: In the formula, , , These are adaptive weighting coefficients that are dynamically adjusted using the tanh function. The rate of change of wave entropy This refers to the acceleration due to load change. It is used to comprehensively reflect the real-time risk level of a system under the combined effects of multiple disturbances. The larger the structure, the more sensitive it is to DER power fluctuations. When the value is >0.8, a 10% change in DER output will cause a voltage deviation of more than 5%; The value ranges from [0,1], with larger values ​​indicating higher risks. Step 3, Tiered Strategy Matching: Based on the chain transmission model, establish a system from power fluctuations ΔP To voltage deviation ΔV From protecting the probability of false activation Pm Predict the system's average outage duration (SAIDI) and average outage frequency (SAIFI); The tiered strategy matching module dynamically triggers control strategies based on risk entropy values ​​and reliability index prediction results. like If the value is >0.8 or the predicted average outage duration SAIDI exceeds 3 hours, a red alert will be activated, and load shedding and energy storage discharge will be performed. If 0.5 < If the value is ≤0.8 or SAIFI exceeds the limit, a yellow alert will be activated, and the microgrid will be switched to islanded operation. In other cases, a blue alert is activated, and SVG reactive power compensation is implemented to achieve multi-level coordinated control that matches the speed of fault evolution. Step 4, Resource Control The resource control layer aims to achieve millisecond-level control command execution and dynamic policy verification to ensure real-time and reliable defense effects. The resource control layer uses a field-programmable gate array (FPGA) as the core control unit. When the hierarchical policy matching module outputs the warning level, the FPGA generates the corresponding control command according to the preset "warning level-command" mapping table, and encapsulates the command code, target device address and priority into a GOOSE message that conforms to the IEC 61850 standard. The instructions are sent to each execution terminal, including the distribution transformer terminal (DTU), energy storage controller, and SVG controller, via the fiber optic ring network with a transmission delay of less than 10ms. The execution terminals parse the messages in real time and trigger the operation. The resource control layer also has a strategy verification and feedback mechanism. By introducing the distribution network PMU to collect system status data in real time, the actual voltage, power and other data are compared with the expected status. If the deviation exceeds the threshold, the strategy parameters are dynamically adjusted and the instructions are regenerated. At the same time, the data of the whole process is recorded in the database for strategy backtracking and optimization, forming a closed-loop control system of "execution-monitoring-verification-adjustment", which greatly improves the control accuracy and robustness of the system in real disturbance environment.

Images