A power distribution network island fault identification and multi-target recovery method and system

Abstract

The application relates to the technical field of active power distribution network fault recovery and protection control, and particularly discloses a power distribution network island fault identification and multi-target recovery method and system, which specifically comprises the following steps: constructing an island fault state model of a photovoltaic power distribution network with extremely high permeability, using a hybrid anti-islanding detection method to judge the operation state of the power distribution network based on the island fault state model, and identifying and confirming the island operation state; based on the confirmed island operation state, a distributed power source and energy storage collaborative control model is constructed; the model comprises a light storage-EV collaborative constraint system; based on the island operation state and the light storage-EV collaborative constraint system, an optimization target of minimizing the power loss load weight, minimizing the number of switch actions and maximizing the EV user satisfaction is determined, and solving is carried out, so that a multi-target island recovery strategy is obtained; the application can accurately identify the island, guarantee the system stability, and take into account the power supply reliability, operation economy and EV user satisfaction.

Description

Technical Field

[0001] This invention relates to the field of active distribution network fault recovery and protection control technology, and in particular to a method and system for identifying islanding faults and multi-target recovery in distribution networks. Background Technology

[0002] The penetration rate of renewable energy sources such as photovoltaics in power distribution networks continues to rise, gradually forming a photovoltaic power distribution network with extremely high penetration. This type of power distribution network not only has the advantages of being clean, low-carbon, and highly efficient in energy utilization, but also effectively alleviates the environmental pressure caused by dependence on traditional fossil fuels, becoming an important component of the new power system.

[0003] However, the large-scale application of photovoltaic distribution networks with extremely high penetration also faces many technical challenges: On the one hand, photovoltaic output is significantly random and volatile due to natural factors such as light intensity, and load demand has time-varying characteristics. The peak electricity consumption periods of different types of loads, such as government / medical, commercial, and residential, are significantly different, which greatly increases the difficulty of power balancing between the source and load of the distribution network. On the other hand, the widespread access of new controllable resources such as electric vehicles (EVs) makes the distribution network exhibit multi-element coupling characteristics of "source-load-storage-vehicle". The charging and discharging behavior of EVs is closely related to the spatiotemporal distribution of traffic flow, which further exacerbates the complexity of distribution network operation.

[0004] In terms of fault recovery, traditional strategies are often geared towards a single objective (such as minimizing power loss load or reducing the number of switching operations), failing to comprehensively consider EV user satisfaction and the rationality of islanding, and failing to effectively coordinate the regulation capabilities of multiple resources such as photovoltaics, energy storage, and EV clusters. In addition, existing recovery algorithms often suffer from slow convergence speed and are prone to getting trapped in local optima, making it difficult to quickly output the optimal recovery solution adapted to high-penetration photovoltaic distribution networks, resulting in delays in power restoration to critical loads and a poor user experience. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for islanding fault identification and multi-objective recovery in distribution networks. By constructing an islanding fault state model that includes photovoltaic, energy storage, and EV clusters, it comprehensively covers the characteristics of multi-source resources, load timing differences, and inverter dynamic characteristics, and clarifies the critical conditions for islanding formation and stable operation boundaries, providing fundamental support for subsequent protection strategy formulation, collaborative control optimization, and recovery scheme solutions.

[0006] On the one hand, a method for islanding fault identification and multi-target recovery in distribution networks is provided, including: Based on the photovoltaic power output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model, an islanding fault state model for a photovoltaic distribution network with extremely high penetration rate is constructed. Based on the islanding fault state model, a hybrid anti-islanding detection method is used to determine the operating state of the distribution network, identify and confirm the islanding operating state; Based on the confirmed islanded operation status, a distributed power source and energy storage collaborative control model is constructed; this model includes: a photovoltaic-energy storage-EV collaborative constraint system. Based on the islanded operation status and the photovoltaic-storage-EV collaborative constraint system, the optimization objectives are determined to be minimizing the power loss load weight, minimizing the number of switching actions, and maximizing EV user satisfaction. The multi-objective islanding recovery strategy is obtained by using dynamic islanding search and improved binary antlion algorithm.

[0007] Preferably, based on the islanded operation status, grid-connected inverter power command constraints and VSC multi-mode smooth switching technology are adopted to protect the islanded system and realize bidirectional smooth switching between islanded operation and grid-connected operation.

[0008] Preferably, the power command constraint of the grid-type inverter is achieved by calculating the active power command constraint value and adjusting the active power command value in combination with staged dynamic constraint logic.

[0009] Preferably, the photovoltaic-storage-EV collaborative constraint system includes islanded power balance constraints, distribution network islanding topology constraints, EV cluster operation constraints, photovoltaic power supply and energy storage operation constraints, and power flow and system security constraints.

[0010] Preferably, the hybrid anti-islanding detection method is a two-stage collaborative strategy of active disturbance injection and passive feature monitoring.

[0011] Preferably, the dual-stage collaborative strategy of active disturbance injection and passive feature monitoring is as follows: a small frequency disturbance is applied to the output power of the grid-connected inverter through active disturbance injection to obtain the detection result; the islanding voltage amplitude, harmonic distortion rate and power fluctuation characteristics are collected simultaneously through passive feature monitoring; when the voltage amplitude deviation exceeds the first preset threshold, the harmonic distortion rate exceeds the second preset threshold or the power fluctuation amplitude exceeds the third preset threshold, cross-validation is performed with the detection result of active disturbance injection to confirm the islanding state.

[0012] Preferably, an EV cluster aggregation charging and discharging model is constructed based on the EV individual emergency charging and discharging model and the traffic flow-electricity flow coupling model.

[0013] On the other hand, a distribution network islanding fault identification and multi-target recovery system is provided, including: The islanding fault state model construction module constructs an islanding fault state model for a photovoltaic distribution network with extremely high penetration rate based on the photovoltaic output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model. The islanding identification and confirmation module, based on the islanding fault state model, uses a hybrid anti-islanding detection method to judge the operating status of the distribution network, identify and confirm the islanding operating status; The collaborative control model construction module constructs a collaborative control model for distributed power sources and energy storage based on the confirmed islanded operation status; this model includes: a photovoltaic-energy storage-EV collaborative constraint system; The multi-objective islanding recovery module, based on the islanding operation status and the photovoltaic-storage-EV collaborative constraint system, determines the optimization objectives as minimizing the power loss load weight, minimizing the number of switching actions, and maximizing EV user satisfaction. It uses dynamic islanding search and an improved binary antlion algorithm to solve the problem and obtain the multi-objective islanding recovery strategy.

[0014] Furthermore, an electronic device is also provided, including: Memory, used for non-transitory storage of computer-readable instructions; and Processor, for executing the computer-readable instructions, When the computer-readable instructions are executed by the processor, they perform the method described in the first aspect above.

[0015] In another aspect, a storage medium is also provided for non-transitory storage of computer-readable instructions, wherein when the non-transitory computer-readable instructions are executed by a computer, the method described in the first aspect is performed.

[0016] The above technical solution has the following advantages or beneficial effects: This invention constructs a three-source collaborative islanding model of photovoltaic-energy storage-EV clusters, which fully considers the strong randomness of photovoltaic output, the time-varying nature of load, and the coupling characteristics of EV clusters and traffic flow under extremely high penetration. At the same time, it quantifies the load recovery priority through weighted load coefficients, accurately characterizes the islanding formation mechanism and stable operation boundary, and provides reliable support for subsequent protection and recovery strategies, greatly improving the adaptability of the technical solution to photovoltaic distribution network scenarios with extremely high penetration.

[0017] This invention employs a hybrid detection method combining active disturbance injection and passive feature detection, effectively avoiding the blind spots, misjudgments, and missed detections inherent in single detection methods in high-penetration scenarios. The islanding detection response is fast and accurate. By combining GFM inverter power command constraints with VSC multi-mode smooth switching technology, it can achieve precise current limiting of fault current and maintain stable power angle, while ensuring seamless connection between the island and the main grid. This completely solves the core contradiction between fault current limiting and mode switching impact in traditional protection strategies, significantly improving the safety of islanding operation. Attached Figure Description

[0018] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0019] Figure 1 This is an overall flowchart of a method for identifying islanding faults and multi-target recovery in power distribution networks. Detailed Implementation

[0020] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the invention. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] In this embodiment of the invention, "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of this invention, "multiple" refers to two or more.

[0023] Furthermore, to facilitate a clear description of the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0024] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0025] Example 1 Extremely high penetration photovoltaic distribution network: refers to a distribution network in which the capacity of distributed power sources such as photovoltaics accounts for no less than 50% of the peak load of the distribution network, and has the operating characteristics of high proportion of power electronic equipment, strong random fluctuation of source and load, and multi-element coupling.

[0026] EV cluster: refers to an aggregation of multiple electric vehicles that can participate in grid interaction and control within the same distribution network area. It has the characteristics of individual emergency charging and discharging and spatiotemporal distribution coupled with traffic flow, and is an important flexible control resource under islanded operation.

[0027] Phase angle synchronization and pre-synchronization control strategy: The phase angle synchronization and pre-synchronization control strategy is a pre-coordinated adjustment control method that coordinates photovoltaic and EV clusters to adjust the voltage amplitude, frequency and phase on the island side to match the main grid before island recovery and grid connection, thereby eliminating the closing inrush current.

[0028] To address the operational characteristics and fault recovery challenges of photovoltaic power distribution networks with extremely high penetration rates, this invention aims to solve the following technical problems: (1) Existing power grid fault models do not fully take into account the strong randomness of photovoltaic output under extremely high penetration, the time-varying nature of load, and the coupling characteristics of EV clusters and traffic flow. They are difficult to accurately depict the island formation mechanism and stable operation boundary, resulting in a lack of reliable modeling support for protection and recovery strategies.

[0029] (2) Traditional single island detection methods are prone to blind spots, misjudgments or omissions in high-penetration photovoltaic scenarios. Furthermore, they do not combine the power angle stability characteristics of GFM inverters with the multi-mode control potential of VSC. When a fault occurs, it is easy to cause problems such as current overload and power angle instability. Secondly, photovoltaic, energy storage and EV clusters have not formed effective coordination, and cannot smooth photovoltaic fluctuations and balance source load power through multi-source complementarity, which further reduces the stability of island operation.

[0030] (3) Traditional recovery strategies focus on a single objective and do not take into account the economic efficiency of switching operations, EV user satisfaction and the rationality of islanding; moreover, the solution algorithm converges slowly and is prone to getting trapped in local optima, making it difficult to quickly output the optimal recovery scheme that is suitable for high penetration scenarios, resulting in delays in the recovery of important loads and poor overall system operation efficiency. At the same time, the existing safety constraint system cannot adapt to the characteristics of complex distribution network topology and increased power fluctuations under high penetration, which can easily lead to safety risks.

[0031] To address key technical challenges such as islanding characteristic modeling, precise protection, PV-energy synergy, and efficient recovery in ultra-high penetration PV distribution networks, it is urgent to propose a PV-energy synergy islanding protection and multi-objective recovery strategy that balances safety, reliability, and economy to support the safe and stable operation of ultra-high penetration PV distribution networks.

[0032] This embodiment provides a method for identifying islanding faults and multi-target recovery in distribution networks, such as... Figure 1 As shown, the method includes: S1: Based on the photovoltaic output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model, construct an islanding fault state model for a photovoltaic distribution network with extremely high penetration rate.

[0033] The islanding fault state model of the ultra-high penetration photovoltaic distribution network couples multiple elements of "source-load-storage-vehicle-grid", providing modeling support for subsequent islanding fault feature identification.

[0034] Specifically: (1) Photovoltaic power output model A photovoltaic power output model is constructed based on the radiation area of ​​photovoltaic modules, solar radiation intensity, and photoelectric conversion efficiency.

[0035] Photovoltaic power output is mainly affected by solar irradiance, which has significant randomness. Therefore, based on the probability distribution characteristics of solar irradiance, a photovoltaic power output model is constructed: (1) in, It is the predicted output value (kW) of photovoltaic power (PVG). It is the photoelectric conversion efficiency, which is determined according to the type of photovoltaic module, and is usually taken as 0.15~0.25; It is the effective radiation area (m²) of the photovoltaic module, which can be calculated based on the actual installed capacity of the photovoltaic power station, the module model and quantity, or it can be obtained by multiplying the area of ​​a typical single module (e.g., 1.6~2.0m² / module) by the number of modules in a simulation scenario; The solar radiation intensity (kW / m²) has a probability distribution that follows a Beta distribution with the following probability density:

[0036] in, It is the Gamma function. , The shape parameters of the Beta distribution are obtained by fitting historical radiation data. The maximum solar radiation intensity (kW / m).

[0037] (2) Weighted Load Model The weighted load model takes into account the time-varying nature and importance differences of the load, and constructs a weighted load model to provide a basis for load priority ranking in island recovery.

[0038] The weighted load model is used for load priority ranking in islanded fault recovery scenarios. It consists of two parts: a time-varying load model and a weighted load coefficient model. First, the load power consumption of different nodes at different times is calculated through the time-varying load model. Then, the weighted load coefficient model is constructed by combining the load level and time period importance to quantify the load recovery priority of each node.

[0039] (2.1) Time-varying load model Distribution network loads are categorized into three types based on industry: government / medical loads, commercial loads, and residential loads. The electricity consumption of these different load types exhibits varying patterns over time. The electricity consumption at each load node in the distribution network at different times is calculated by integrating a typical daily load curve. (3) in, It is a node exist Electricity consumption at any given time (kWh); It is a node The corresponding typical daily load variation curve (kW) reflects the load variation trend over time.

[0040] (2.2) Weighted Load Factor Model Based on the degree of power loss caused by load failure, loads are classified into three levels: Level 1, Level 2, and Level 3. Level 1 loads are critical loads, such as government and medical loads; Level 2 loads are less critical loads, such as commercial loads; and Level 3 loads are general loads, such as some residential loads. A weighted load factor is defined by combining the load level with the difference in importance at different times. To quantify the load recovery priority of each node: (4) in, It is the weight load coefficient of node i at time t. The larger the value, the higher the recovery priority. It is the load recovery coefficient at time t. This refers to the difference in importance at different times. The recovery coefficient for the primary load is set to the maximum value, while other load levels are set according to the importance of the fault at the time of the fault. For example, when a fault occurs in the morning or evening, the residential load is more important, so the residential load is set to 6 and the commercial load to 1; while when a fault occurs during working hours, the government and medical loads are more important, so the government and medical loads are set to 6 and the residential load to 1. This is the load level factor; for level I, II, and III loads, Take values ​​of 1, 0.1, and 0.01 respectively; It is the electricity consumption (kWh) of node i at time t, which is calculated by the time-varying load model.

[0041] (3) Emergency charging and discharging model of EV cells To address the core objective of "prioritizing power supply to critical loads while also considering the basic travel needs of EV users" in islanded fault scenarios, an emergency charging and discharging model for individual EV units is established. This model clarifies the boundary constraints on electrical energy, dynamic constraints on charging and discharging power, and response constraints for islanded mode switching, as detailed below: (3.1) Electrical energy boundary constraints: (5) in, , These are the maximum and minimum allowable electrical energy (kWh) of the i-th EV at time t, respectively. It is the initial electrical energy (kWh) of the i-th EV at the initial time t0 when it is connected to the grid. , These are the maximum and minimum charging / discharging power (kW) of the i-th EV, respectively, with positive values ​​during discharging and negative values ​​during charging; It is the initial moment (h) when the EV is connected to the grid. It is the maximum electrical energy (kWh) of the battery of the i-th EV, that is, the rated capacity of the battery; This is the emergency reserve power (kWh) for the i-th EV in an isolated scenario, used to ensure the user's necessary travel needs, and is taken as 20% to 30% of the battery's rated capacity; It is the battery protection energy (kWh) of the i-th EV, which is used to prevent overcharging and discharging from damaging the battery. The value is 5% to 10% of the battery's rated capacity.

[0042] (3.2) Dynamic constraints on charge and discharge power: (6) in, , These are the maximum and minimum allowable charging / discharging power (kW) of the i-th EV at time t, with discharging being positive and charging being negative; , These are the maximum and minimum allowable electric energy (kWh) of the i-th EV at time t+1, respectively. It is the power surplus (kW) of the photovoltaic-energy storage combined system in the island at time t, that is, the remaining power after subtracting the total load demand in the island from the total output of the combined system; It is the power deficit (kW) within the island at time t, which is the difference between the total load demand within the island and the total output of the "photovoltaic-energy storage" combined system.

[0043] (3.3) Island mode switching response constraints: (7) in, , These are the i-th EV at time t and ti respectively. Actual charging / discharging power (kW) at moment 1; It is the maximum rate of change of charging and discharging power of the i-th EV (kW / min), which is set according to the response speed of the EV inverter and is usually taken as 5~10kW / min.

[0044] (4) Traffic flow-electric flow coupling model Considering that traffic flow is affected by faults in the island scenario and its spatiotemporal distribution changes, and that the EV grid connection status is closely related to traffic flow, a traffic flow-electric flow coupling model is established to achieve temporal linkage and collaborative optimization between the two.

[0045] The dynamic traffic flow constraints of the traffic flow-electricity flow coupled model are divided into four levels from the basic to the coupled, with the following logical relationship: Traffic arc flow balance equation: as a fundamental constraint, it describes the dynamic flow of vehicles in the road network; Traffic flow constraints in island-affected areas: Based on basic traffic flow, the limitations on road network capacity caused by island failures are introduced to modify the upper limit of vehicle flow. EV grid-connected status time-series mapping: Transform the traffic flow distribution in the island area into an EV cluster scale that can participate in grid regulation, and realize the time-series linkage between traffic flow and power flow; Regional coordination constraints: Imposing distribution network capacity limits on EV clusters participating in regulation to ensure that EV charging and discharging behavior does not exceed the safe operating boundaries of the islanded distribution network. Specifically: (4.1) The traffic flow balance equation is as follows: (8) (9) in, It is the vehicle flow (in vehicles) of traffic arc a at time t. A traffic arc refers to the road segment between adjacent intersections. , These are the vehicle flow rates (vehicles / h) flowing into and out of traffic arc a at time t. D(j) is the set of traffic arcs starting from intersection j; D(j) is the set of traffic arcs ending at intersection j.

[0046] (4.2) Traffic flow constraints in the island's impact area: (10) in, It is the maximum capacity (vehicles) of traffic arc a. It is the islanding impact coefficient, which is 0.3~0.5 in the initial stage of the fault (0~1h) and 0.8~1.0 in the later stage of recovery (4~6h).

[0047] (4.3) EV grid-connected state timing mapping: (11) in, It is the size (vehicles) of the EV cluster that can participate in islanded control at time t. It is the set of traffic arcs within the coverage area of ​​the isolated island; This represents the percentage of EVs on traffic arc a, using historical statistics ranging from 0.1% to 0.3%. This is the EV grid connection willingness coefficient, which is 0.8~1.0 in emergency situations and 0.5~0.8 in normal recovery phases.

[0048] (4.4) Regional Coordination Constraints: (12) in, It is the set of EVs within the k-th isolated partition; It is the actual charging / discharging power (kW) of the i-th EV at time t. It is the distribution network carrying capacity (kW) of the kth isolated zone, calculated based on line impedance and transformer capacity.

[0049] (5) EV cluster aggregation charge and discharge model Based on the emergency charging and discharging model of individual EVs and the traffic flow-power flow coupling model, an aggregated charging and discharging model of EV clusters is constructed to achieve coordinated control with the "photovoltaic-energy storage" system, providing overall support for island power balance.

[0050] This model is achieved through the coordinated implementation of three parts: total charging power of the cluster, total discharging power of the cluster, and state of charge (SOC) constraints of the cluster. Total charging / discharging power: Based on the charging / discharging power and access status of individual EVs, the power interaction capability of the entire cluster is aggregated and directly used as the input term of the island power balance equation; SOC constraints: Set safety boundaries for cluster charging and discharging behavior to avoid overcharging and discharging while ensuring users' basic travel needs.

[0051] (5.1) Total charging power of the cluster: (13) in, ζ(t) is the total charging power of the EV cluster at time t (kW); ζ(t) is the island power balance coefficient. This refers to EV charging efficiency, typically between 0.9 and 0.95. It is the access status identifier for the i-th EV. When not connected 0; It is the single-unit charging power (kW) of the i-th EV at time t.

[0052] (5.2) Total discharge power of the cluster: (14) in, It is the total discharge power of the EV cluster at time t (kW); This refers to the EV discharge efficiency, which is taken as 0.9~0.95. It is the single-unit discharge power (kW) of the i-th EV at time t. It is the discharge status indicator for the i-th EV, which is 1 when it can discharge and 0 when it cannot discharge.

[0053] (5.3) Cluster State of Charge (SOC) Constraints:

[0054] Among them, 0.3 and 0.8 are the minimum and maximum allowable states of charge for the EV cluster, respectively; It is the actual electrical energy (kWh) of the i-th EV at time t.

[0055] (6) Transient model of grid-connected system with GFM inverter A transient model of a grid-connected GFM inverter system is constructed based on a droop control transient model. The output power of the grid-connected inverter is output, the system stability characteristics under voltage drop conditions are analyzed, and the critical conditions for islanding formation and stable operation are clarified.

[0056] GFM power control utilizes the strong coupling relationship between active power and frequency, and reactive power and voltage, to establish active power droop control equations and reactive power droop control equations for the GFM inverter. These two different core control equations are shown below:

[0057]

[0058] in, , These are the reference values ​​for the grid angular frequency and voltage, respectively. , These are the power loop output angular frequency and the power loop output voltage, respectively. , These are the command values ​​for active power and reactive power, respectively. , These are the active and reactive power droop coefficients, respectively.

[0059] The expressions for P and Q are as follows:

[0060]

[0061] in:

[0062] in, , These are the grid voltage and the line equivalent inductance, respectively. The power angle is the phase difference between the inverter output voltage and the grid voltage. The equivalent reference voltage; Line susceptance (S); Generation of the droop control transient model: By simultaneously solving the active power droop control equations and reactive power droop control equations of the GFM inverter and the output power expression of the grid-connected inverter, the formula for the droop control transient model can be obtained as follows:

[0063]

[0064] Among them, the droop control transient model uses the power angle δ and the output voltage amplitude U as state variables, which can further derive the angular frequency change rate and characterize the transient response characteristics of the inverter under islanding faults.

[0065] The rate of change of angular frequency is further derived from the droop-controlled transient model, specifically as follows: (twenty three) in, The angular frequency change rate is rad / s².

[0066] Finally, by combining the photovoltaic output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model, an islanding fault state model of ultra-high penetration photovoltaic distribution network is constructed. This model clarifies the coupling relationship and operating characteristics of each component under fault conditions, providing a foundation for islanding fault feature identification.

[0067] The islanding fault state model of the ultra-high penetration photovoltaic distribution network constructed in this invention can fully take into account the strong randomness of photovoltaic output, the time-varying nature of load, and the coupling characteristics of EV clusters and traffic flow under ultra-high penetration, accurately depict the islanding formation mechanism and stable operation boundary, and provide reliable modeling support for protection and recovery strategies.

[0068] The islanding fault state model constructed in this invention uses "photovoltaic-energy storage-EV-GFM inverter-distribution network load" as the core coupling unit, covering three core modules: power balance, component constraints, and timing linkage under fault conditions. The core power balance expression of the islanding fault state model integrates the outputs of all the aforementioned sub-models, achieving a unified characterization of the power flow across the entire islanding process; the specific expression is as follows:

[0069] in, It is the actual output (kW) of the i-th photovoltaic node at time t; , These are the charging and discharging power (kW) of the energy storage device at time t, respectively. , These represent the total discharge and charging power (kW) of the EV cluster at time t, respectively. The total load power (kW) of the distribution network at time t; The power loss (kW) of the distribution network line at time t is calculated based on the line impedance and current, as shown below:

[0070] in, , These represent the current and resistance of line ij, respectively.

[0071] Furthermore, the islanded fault state model constructed in this invention must satisfy the following constraints: Photovoltaic output constraints:

[0072] in, , These are the minimum and maximum output (kW) of the i-th photovoltaic node at time t, respectively, determined by the Beta distribution of irradiance and the characteristics of the module.

[0073] Energy storage operation constraints:

[0074]

[0075]

[0076]

[0077] in, , The charge / discharge state coefficient, ∈{0,1}; The energy storage state of charge at time t; , These represent the minimum and maximum states of charge, respectively, and are taken as 0.2~0.8.

[0078] Distribution network security constraints:

[0079]

[0080] Inverter operation constraints: satisfy the GFM inverter transient constraints of formulas (16) to (23) to ensure that the inverter output is stable under fault conditions.

[0081] Based on the above islanding fault state model, and combined with the transient characteristics of the GFM inverter and the coupling relationship of each component, the core state characteristics of islanding faults are systematically analyzed, and the key criteria for islanding formation, stable operation and instability are clarified, providing a basis for the formulation of islanding fault identification and recovery strategies.

[0082] Islanding characteristics: When a grid fault causes the system voltage to drop below the critical voltage, the system experiences power angle instability. The original stable and unstable equilibrium points disappear, and the grid-connected inverter switches from grid-connected control mode to islanded voltage regulation control mode. Photovoltaics, energy storage, and electric vehicle clusters collectively form an independent power supply unit. During the islanding phase, the system gradually transitions from a transient state to a steady state. Within the island, power balance is maintained through the coordinated efforts of various power sources and loads, achieving independent operation after electrical isolation from the main grid.

[0083] Islanded stable operation characteristics: After entering steady-state islanded operation, the system angular frequency remains constant, the node voltage remains within the allowable fluctuation range of the rated voltage, the state of charge of the energy storage device is within the safe operating range, and the average state of charge of the electric vehicle cluster remains at a reasonable level. Random fluctuations in photovoltaic output are jointly mitigated by energy storage and the electric vehicle cluster, there are no frequent operations of the distribution network switching equipment, the load power supply priority is executed in an orderly manner according to the preset weight coefficient, and the overall system operation is stable and controllable.

[0084] Islanding instability characteristics: An islanded system is considered unstable when any of the following conditions are met: 1) Node voltage exceeds the allowable fluctuation range and persists for more than the specified duration; 2) Energy storage state of charge exceeds the safe operating range; 3) The average state of charge of the electric vehicle cluster is too low and there is no other power supply; 4) The inverter power angle exceeds the stable operating range; 5) Line current exceeds the allowable current carrying capacity and persists for more than the specified duration. The occurrence of any of the above situations indicates that the islanding system cannot maintain safe and stable operation, and control measures such as load shedding and adjusting the islanding range must be taken promptly.

[0085] Fault state differentiation characteristics: Minor fault: The photovoltaic output is basically stable, and only minor adjustments to the energy storage are needed to maintain power balance. No islanding has been formed, and the GFM inverter maintains grid-connected control mode. Moderate fault: The initial islanding has been formed, and the output of photovoltaic power has fluctuated more. It is necessary to coordinate the regulation of energy storage and EV clusters. The GFM inverter switches to islanding control mode, and the voltage and frequency are basically stable. Severe fault: The island is completely independent, and the photovoltaic output is highly random. It is necessary to strictly control the active power command constraints. The EV cluster is given priority to discharge to make up for the power deficit. The energy storage state of charge drops rapidly, and the island range needs to be dynamically adjusted to maintain stability.

[0086] Evolution of Equilibrium Point and Stability Boundary under Voltage Dip: To clarify the dynamic characteristics of the system during voltage dip, based on the transient model of the grid-connected inverter and the power balance relationship, the evolution law and stability boundary of the system equilibrium point are analyzed through theoretical derivation, providing a theoretical basis for the formulation of islanding protection and fault recovery strategies.

[0087] The core criterion for steady-state operation of the system is that the angular frequency deviation Δω=0. Under this condition, the intersection of the system state surface and the Δω=0 stable plane is the system equilibrium point, which is specifically divided into the stable equilibrium point (SEP) and the unstable equilibrium point (UEP). The two together define the stability boundary for the normal operation of the grid-connected system.

[0088] Under steady-state conditions before the power grid fault, the grid voltage Ug = (Rated voltage), at this time, the Δω=0 plane and the system state surface have two clear and independent intersection points, namely SEP and UEP. The system is in a stable grid-connected operation state, and the position distribution of the two points determines the system's disturbance rejection capability.

[0089] When a grid fault causes Ug to gradually decrease, SEP tends to shift towards increasing power angle δ, while UEP moves towards decreasing δ. When both converge to the same critical point (BP), the system has only a single equilibrium operating point, and the grid-connected system is in a critical stable state. When the grid voltage drop exceeds the voltage threshold corresponding to BP... When the Δω=0 plane completely loses its intersection with the system state surface, both SEP and UEP disappear, and the system experiences power angle instability. At this time, islanding formation or fault ride-through strategies need to be triggered.

[0090] Setting Δω=0, and combining the transient model of the grid-connected inverter with the power constraint conditions, the critical voltage can be derived. :

[0091] By substituting the system parameters into the above formula, the critical voltage value for stable islanding operation under different operating conditions can be determined, providing a basis for the formulation of islanding protection and recovery strategies.

[0092] S2: Based on the islanding fault state model, a hybrid anti-islanding detection method is used to determine the operating status of the distribution network, identify and confirm the islanding operating status.

[0093] Based on the islanded operation status, grid-connected inverter power command constraints and VSC multi-mode smooth switching technology are adopted to protect the islanded system and realize bidirectional smooth switching between islanded operation and grid-connected operation. The grid-connected inverter power command constraints are realized by calculating the active power command constraint value and adjusting the active power command value in combination with staged dynamic constraint logic.

[0094] By integrating active disturbance injection and passive feature detection into a hybrid anti-islanding detection method, and combining grid-connected inverter power command constraints with VSC multi-mode smooth switching technology, the system achieves accurate islanding identification, fault current limiting, and smooth transition from islanding to the main grid, comprehensively ensuring the operational safety and stability of photovoltaic distribution networks with extremely high penetration rates during faults.

[0095] To address the issue of blind spots in single anti-islanding detection methods in high-penetration photovoltaic access scenarios, a dual-stage collaborative strategy of active perturbation injection and passive feature detection is adopted to improve the accuracy, reliability, and speed of detection.

[0096] The hybrid anti-islanding detection method employs a dual-stage collaborative strategy of active disturbance injection and passive feature monitoring. This strategy works as follows: A small frequency disturbance is applied to the output power of the grid-connected inverter via active disturbance injection to obtain the detection result; simultaneously, islanding voltage amplitude, harmonic distortion rate, and power fluctuation characteristics are acquired through passive feature monitoring; when the detected voltage amplitude deviation exceeds a first preset threshold, harmonic distortion rate exceeds a second preset threshold, or power fluctuation amplitude exceeds a third preset threshold, cross-validation is performed with the detection result of the active disturbance injection to confirm the islanding status.

[0097] The active disturbance injection process involves applying a small frequency disturbance to the output power of the grid-connected inverter in the transient model of the GFM inverter grid-connected system, and monitoring the islanding voltage frequency offset response in real time. The disturbance amplitude is strictly controlled to ≤0.2Hz to avoid affecting normal power supply quality. If the frequency offset exceeds ±0.5Hz for three consecutive power frequency cycles (≤60ms), an islanding warning signal is immediately triggered, providing a preliminary judgment for subsequent protection actions. Here, the frequency offset is the difference between the voltage frequency after disturbance injection and the rated frequency.

[0098] The passive feature monitoring stage involves synchronously acquiring key parameters such as islanded voltage amplitude, harmonic distortion rate, and power fluctuation characteristics. When a voltage amplitude deviation exceeding ±5%, a harmonic distortion rate exceeding 5%, or a power fluctuation amplitude exceeding 20% ​​is detected, and the duration is ≥0.1s, the results are cross-validated with the detection results from the active disturbance injection stage to confirm the islanded status.

[0099] By coordinating the active disturbance injection stage and the passive feature monitoring stage, the risk of misjudgment and missed judgment of a single detection method is effectively avoided. The island detection response time is ≤20ms, which meets the speed requirements of island detection for photovoltaic power distribution networks with extremely high penetration.

[0100] Furthermore, based on the islanded operation status, grid-connected inverter power command constraints and VSC multi-mode smooth switching technology are adopted to protect the islanded system and perform bidirectional smooth switching between islanded and grid-connected operation. The grid-connected inverter power command constraints are achieved by calculating the active power command constraint value and adjusting the active power command value in combination with staged dynamic constraint logic.

[0101] To address the issues of power angle instability and output current overload that are prone to occur in grid-connected inverters during symmetrical grid faults, a power command constraint strategy for grid-connected inverters is proposed. This strategy is based on a fault ride-through strategy with active power command constraints. It does not require additional current limiting modules or virtual impedance links. It can achieve stable reshaping of the power angle and precise limitation of fault current simply by optimizing the active power command value, while preserving the grid-connected inverter's ability to support the voltage and frequency of the system.

[0102] The power command constraint objective of grid-connected inverters is to ensure the inverter's power angle remains stable (maintaining the stable equilibrium point SEP) under grid voltage dip conditions, while limiting the fault current amplitude to within 1.5 times the rated current to prevent the inverter from disconnecting from the grid due to overcurrent.

[0103] The core principle is as follows: When the grid voltage drops, the coupling relationship between the inverter output power and the power angle changes, resulting in no intersection between the original active power command value and the power angle characteristic curve (P-δ curve), and the disappearance of SEP causes power angle instability. This invention can re-establish the intersection between the active power command and the P-δ curve by reasonably reducing the active power command value, restore SEP, and at the same time reduce the fault current amplitude.

[0104] The calculation process for the active power command constraint value is as follows: Based on the transient model of the grid-type inverter, and combined with the constraint condition mentioned above that the fault current amplitude is limited to within 1.5 times the rated current, the active power command constraint value is derived. The calculation formula is:

[0105] in, The active power command constraint value (kW) under fault conditions; This is the reactive power droop factor, which is set according to the system voltage regulation requirements. The grid fault voltage (kV) is obtained in real time through inverter detection. Line susceptance (S) Where Lg is the equivalent inductance of the line (mH). The rated angular frequency of the power grid (rad / s); The critical power angle (rad) corresponding to the fault current limit is derived from the fault current constraint condition. The rated voltage of the power grid is (kV). The reactive power command value (kvar) remains unchanged during a fault.

[0106] Furthermore, to adapt to real-time changes in grid fault voltage, a phased dynamic constraint logic is adopted to adjust the active power command constraint value, ensuring system stability under different voltage drop levels.

[0107] The staged dynamic constraint logic is as follows: Real-time detection of grid voltage sag coefficient ; Mild voltage dip (dip factor ≥ 0.5): Maintain the original active power command value and provide transient support only through VSC control mode switching to ensure system stability; Moderate voltage sag (0.2 ≤ sag coefficient < 0.5): Initiate active power command constraints and calculate the active power command constraint value using the formula above. The active power command of the inverter is updated to the calculated value, while the rate of change of the command value is limited to ≤0.1kW / ms to avoid secondary impact caused by power surges. Severe voltage drop (drop factor < 0.2): Further reduce the active power command constraint value to ensure that the fault current does not exceed 1.5 times the rated current, meet the equipment overload capacity requirements, and maintain stable operation of the inverter power angle. The active power command constraint value is calculated based on the transient model of the grid-type inverter and the fault current constraint condition.

[0108] Furthermore, the VSC multi-mode smooth switching technology utilizes the multi-mode control characteristics of the voltage source converter (VSC) and combines phase angle synchronization and pre-synchronization control strategies to achieve a smooth transition between islanded and main grid, avoid voltage surges and power fluctuations caused by mode switching, and ensure power supply continuity.

[0109] VSC multi-mode control characteristics: For AC / DC hybrid networks, the VSC converter can select two variables for control, such as active / reactive power and AC / DC voltage. Specifically, there are three control modes: (1) Control. Controls DC voltage and reactive power to maintain a constant DC line voltage; suitable for DC side voltage stability control during normal operation. (2) Control. Controls active and reactive power to achieve flexible power exchange with the AC side, suitable for normal grid-connected operation scenarios; (3) Control. Controls the amplitude and phase angle of AC voltage to provide voltage support and frequency stability to passive AC networks, suitable for islanded operation scenarios.

[0110] Utilizing the multi-mode control characteristics of a voltage source converter (VSC), and combining phase angle synchronization and pre-synchronization control strategies, the mode switching logic is implemented. The process is as follows: Under normal operating conditions and fault recovery conditions, the VSC (Variable Storage and Control Center) adopts different operating modes in the distribution network. During normal operation, the VSC station uses... Control. When a fault occurs on the AC side, the AC-side loads that have lost power form a passive AC network. At this time, the VSC (Voltage Control Center) is responsible for power transmission, voltage support, and frequency stabilization of the AC network on the lost side. It acts as the power source for the passive AC network. Switch to The control unit acts as the power source for the passive AC network, responsible for power transmission, voltage support, and frequency stabilization. Switching is constrained by phase angle integrals.

[0111] in, For the phase angle before switching (rad); ω is the angular frequency adjustment coefficient (rad / s); s is a complex frequency variable to ensure smooth phase angle transition and voltage surge ≤5%.

[0112] Once the main grid fault is cleared, the VSC initiates pre-synchronization control, adjusting the amplitude and phase of the output voltage to achieve synchronization with the main grid voltage (phase difference ≤ 3°), and then smoothly switches back to P. Q control mode ensures that the grid-connected inrush current is ≤0.3 times the rated current, achieving seamless connection between the island and the main grid.

[0113] This invention employs a hybrid detection method combining active disturbance injection and passive feature detection, effectively avoiding the blind spots, misjudgments, and missed detections inherent in single detection methods in high-penetration scenarios. The islanding detection response is fast and accurate. By combining GFM inverter power command constraints with VSC multi-mode smooth switching technology, it can achieve precise current limiting of fault current and maintain stable power angle, while ensuring seamless connection between the island and the main grid. This completely solves the core contradiction between fault current limiting and mode switching impact in traditional protection strategies, significantly improving the safety of islanding operation.

[0114] Furthermore, this invention constructs a photovoltaic power source and energy storage collaborative control model to achieve power balance and voltage / frequency support in islanded scenarios. The model includes two parts: photovoltaic power source output constraints and distribution network energy storage charging and discharging model.

[0115] The operation of photovoltaic power sources within an isolated island must meet constraints on active and reactive power output and capacity to ensure stable output without exceeding equipment operating limits. Specific output constraints for photovoltaic power sources are as follows:

[0116]

[0117]

[0118] in, , These represent the active and reactive power outputs (kW and kvar) of the photovoltaic power source at node i in time period t, respectively. , These are the minimum and maximum active power output (kW) of the photovoltaic power source at node i during time period t, respectively, calculated using the Beta distribution model of irradiance. Let be the rated capacity (kVA) of the photovoltaic power source at node i. The minimum power factor of wind and solar power at node i is typically taken as 0.85 (lagging).

[0119] Energy storage devices serve as the primary power source for short-term power supply in islanded distribution networks, and their stable output ensures their ability to maintain island voltage and frequency. However, due to their limited capacity, they need to work in conjunction with photovoltaic power sources to mitigate power output fluctuations. The energy storage charging and discharging model for distribution networks is as follows:

[0120]

[0121]

[0122]

[0123] in, , These represent the charging and discharging power (kW) of the energy storage device during time period t. , These are the charge and discharge state coefficients of the energy storage device (0 indicates no operation, 1 indicates operation); The maximum charging / discharging power (kW) of the energy storage device; The state of charge (SOC) of the energy storage device during time period t; This represents the initial state of charge of the energy storage device. The time step is (h), usually taken as 0.25h.

[0124] Furthermore, combining VSC control mode switching and grid-connected inverter power command constraints, an improved binary antlion algorithm is used to achieve fault recovery and reconfiguration, as follows: 1) After islanding is completed, update real-time data such as distribution network topology, load, and power output; 2) Initialize the algorithm parameters; 3) Construct population coding and initialize the antlion and ant populations; 4) Enter iterative optimization to sort the antlion population and obtain elite antlions; 5) By combining chaotic mapping with roulette wheel, the parent antlion ants will walk randomly around the selected antlion; 6) Updated the antlion population and elite antlion collection, and optimized the search direction; 7) Determine if the maximum number of iterations has been reached. If it has, output the optimal fault recovery scheme and the corresponding VSC control mode configuration; if not, return to step 5 to continue iterating. 8) Perform the switch operation and VSC control mode switching according to the optimal solution to complete the fault recovery.

[0125] S3: Based on the confirmed islanded operation status, construct a distributed power source and energy storage collaborative control model; this model includes: photovoltaic-energy storage-EV collaborative constraint system.

[0126] The photovoltaic-storage-EV collaborative constraint system includes islanded power balance constraints, distribution network islanding topology constraints, EV cluster operation constraints, photovoltaic power supply and energy storage operation constraints, and power flow and system security constraints.

[0127] To ensure the safety and stability of the coordinated operation of photovoltaic, energy storage, and EV clusters within an island, this invention constructs a distributed power source and energy storage coordinated control model. This model is used to establish a unified constraint system for photovoltaic, energy storage, and EV clusters in islanded scenarios, ensuring the safety and stability of multi-source coordinated operation. Its specific contents are as follows: Determining the photovoltaic-energy storage-EV coordinated constraint system, which includes islanded power balance constraints, distribution network islanding topology constraints, EV cluster operation constraints, photovoltaic power source and energy storage operation constraints, and power flow and system security constraints.

[0128] At any point during islanded operation of the distribution network, it is necessary to maintain power balance between power supply and load. The islanded power balance constraint is as follows:

[0129] Where n is the number of photovoltaic power nodes in the island; It is the active power output (kW) of the photovoltaic power source at node i during time period t. , These represent the charging and discharging power (kW) of the energy storage device during time period t. , These represent the total discharge and charging power (kW) of the EV cluster during time period t. The total load power (kW) within the island during time period t.

[0130] The topology constraints for islanding distribution networks include node affiliation constraints and line-topology connectivity constraints.

[0131] Each load node in a distribution network can only be contained within one of its islands. The node affiliation constraint is expressed as: (44) in, Let i be the island affiliation status of node i, where 1 indicates that it belongs to island s and 0 indicates that it does not belong to island s. , These are the island set and the distribution network load node set, respectively.

[0132] Line and topology connectivity constraints: After dividing into islands, the topology should not only satisfy connectivity but also maintain a radial shape. Linearizing this constraint yields the following constraints: (45) (46) (47) (48) in, The island destination status of line ij (1 indicates that it is included in island s, and 0 indicates that it is not included). A collection of distribution network lines; This represents the switch state of line ij (1 indicates closed, 0 indicates open). This represents the total number of load nodes in the distribution network. This represents the number of isolated islands formed after the division.

[0133] EV cluster charging and discharging must meet power and state of charge constraints, while balancing grid demand and user travel needs. The specific operational constraints of EV clusters are as follows:

[0134]

[0135] in, Let represent the charging / discharging power (kW) of the i-th EV during time period t, with discharging being positive and charging being negative. , These are the maximum and minimum charging / discharging power (kW) of the i-th EV during time period t; Let be the initial electric energy (kWh) of the i-th EV. The minimum allowable electrical energy (kWh) for EVs is taken as 5% to 10% of the rated capacity; Let be the actual electric energy (kWh) of the i-th EV during time period t. , These are the maximum and minimum allowable electric energy (kWh) of the i-th EV in time period t.

[0136] The operational constraints of photovoltaic power sources and energy storage include the operational constraints of photovoltaic power sources and the operational constraints of energy storage devices, specifically: Operating constraints of photovoltaic power sources:

[0137] Energy storage device operating constraints:

[0138] Power flow and system safety constraints include node voltage constraints, line current constraints, and improved Distflow power flow constraints.

[0139] Node voltage constraints: Node voltages must be maintained within allowable ranges to ensure power quality.

[0140] in, Let be the voltage amplitude (kV) at node i during time period t. , These are the upper and lower voltage limits (kV) of node i, typically taken as ±10% of the rated voltage.

[0141] Line current constraint: The line transmission current shall not exceed the maximum allowable value to avoid overload and burnout.

[0142] in, Let be the current amplitude (kA) of line ij during time period t; Let be the maximum allowable current (kA) for line ij.

[0143] Improved Distflow Power Flow Constraints: Based on the characteristics of radial distribution networks, improved distflow power flow constraints are established to accurately describe power transmission relationships. (55) (56) (57) (58) (59) (60) in, , These represent the active and reactive power (kW, kvar) from node j to sub-node k during time period t. , These are the active and reactive power (kW, kvar) of line ij during time period t, respectively. , Hereinafter, the resistance and reactance (Ω) of line ij are respectively. Let t be the current (kA) of line ij during time period t. , These represent the active and reactive power (kW, kvar) injected into node j during time period t. , These represent the active and reactive power outputs (kW, kvar) of the photovoltaic power source at node j during time period t. , These represent the charging and discharging power (kW) of the energy storage device at node j during time period t. The reactive power output (kvar) of the energy storage device at node j during time period t; , These are the active and reactive discharge power (kW, kvar) of the EV cluster at node j during time period t. , These are the active and reactive loads (kW, kvar) of node j in time period t, respectively. , These are the sets of parent and child nodes of node j in the distribution network, respectively. is the voltage loss correction factor (kV) for line ij.

[0144] The aforementioned collaborative control model is a non-convex nonlinear model. A convex relaxation method is used to relax the constraints, and a second-order cone transformation is employed to convert it into a second-order cone programming model, reducing the solution complexity. Based on the MATLAB R2018a programming environment, the Cplex+yalmip optimization solver is used to solve the strategy, quickly obtaining the optimal operating parameters for the photovoltaic, energy storage, and EV clusters, ensuring the efficiency and stability of the photovoltaic-energy storage-EV collaborative operation.

[0145] S4: Based on the islanded operation status and the photovoltaic-storage-EV collaborative constraint system, the optimization objectives are determined to be minimizing the power loss load weight, minimizing the number of switching actions, and maximizing EV user satisfaction. The multi-objective islanding recovery strategy is obtained by using dynamic island search and improved binary antlion algorithm.

[0146] With the core objectives of minimizing the weight of power loss loads, minimizing the number of switching actions, and maximizing EV user satisfaction, a multi-objective islanding recovery strategy is constructed by integrating dynamic islanding search and an improved binary antlion algorithm. Through a standardized recovery process, it achieves rapid, safe, and efficient recovery of photovoltaic distribution networks with extremely high penetration rates after a fault.

[0147] The process of constructing a multi-objective optimization function is as follows: Taking into account power supply reliability, operational economy, and user satisfaction, a multi-objective optimization function is established to transform the complex recovery problem into a single-objective optimization solution, ensuring the comprehensiveness and rationality of the decision-making process. Specifically:

[0148] The following is the objective function decomposition process: Objective 1: Minimize the weight of power outage loads Optimize and ensure power supply to primary loads, and quantify the overall losses caused by power outages.

[0149] in, The total number of distribution network nodes (including AC and DC nodes); Let be the weight load coefficient of node i at time t; Let be the total load power (kWh) of node i at time t. The power supply status of node i at time t (1 indicates power failure, 0 indicates power supply).

[0150] Objective 2: Minimize the number of switching actions Reduce switching operation costs and equipment wear and tear, and improve recovery efficiency.

[0151] in, The sum of the number of switches, including AC switches and DC switches; This indicates the operating status of AC or DC switch b (1 indicates operation, 0 indicates no operation).

[0152] Objective 3: Maximize EV user satisfaction Using the state of charge (SOC) of EVs when they leave the grid as the core indicator, we aim to avoid impacting users' travel needs.

[0153] in, The total number of EVs participating in the regulation; Let be the electrical energy (kWh) of the i-th EV when it leaves the grid. Let be the maximum energy capacity (kWh) of the i-th EV.

[0154] Objective 4: Optimization of Distribution Network Islanding To reduce the power outage rate after islanding, maximize the value of islanded power supply, and simultaneously consider the power satisfaction of EV users:

[0155] in, For the set of periods of time during which the fault lasts; This refers to the set of load nodes in the distribution network (including EV loads). The importance weight of load node i is set as follows (100 for level 1 load, 10 for level 2 load, and 1 for level 3 load). The load removal status of node i in time period t (1 indicates removal, 0 indicates retention); Let t represent the load size of node i during time period t (kWh). This is the penalty coefficient (ranging from 0.3 to 0.5). Let be the electric energy (kWh) of the i-th EV during time period t. Let be the maximum energy capacity (kWh) of the i-th EV.

[0156] Target weight coefficient setting: (Emphasizing the minimization of power outage load weight) (Taking into account the economic efficiency of switch operation) 15 (Ensuring user satisfaction) and meeting multi-objective equilibrium optimization requirements. (Strengthen the optimization of island partitioning) to achieve multi-objective balanced optimization.

[0157] Constraints for EV energy calculation:

[0158]

[0159] in, The initial electrical energy (kWh) of the i-th EV when connected to the grid. This represents the charging / discharging power (discharging is positive, charging is negative). Power loss during EV operation (kW); EV operating loss rate; The minimum permissible electrical energy for an EV (kWh).

[0160] Furthermore, considering the fluctuations in photovoltaic power output and the uncertainty of faults, and combining the optimization requirements of the islanding objective function, a dynamic islanding search method that prioritizes important loads and dynamically expands is adopted to ensure that the islanding range is dynamically matched with power output, load demand, and EV control capabilities.

[0161] The dynamic islanding search method prioritizes primary critical loads (such as government and medical loads) as the primary search targets. Under the premise of meeting power balance constraints and islanding objectives, it gradually expands the islanding scope, giving priority to restoring critical loads and secondary loads connected to photovoltaic power sources. When there is surplus photovoltaic output, secondary and tertiary loads are further included to maximize the value of load restoration and EV user satisfaction.

[0162] The dynamic island search method is a time-segmented search process, which includes the following: 1) Initialize parameters: Obtain the current period's photovoltaic output forecast, load power data (controllable load reduced to the lowest adjustable range), initial state of charge of energy storage, grid connection status of EV cluster and initial electrical energy, and set the photovoltaic power node as the search root node; 2) Branch Depth-First Search: Search from the root node towards each branch leaf node, traversing all nodes on each branch. The search stops when the following condition is met: If the node is an important load node, and the sum of all loads on the current search path and the loads already included in the island is less than or equal to the total output of photovoltaic + energy storage + EV discharge within the island, proceed to the next verification step; If a node is a critical load but does not meet power balance, or is a leaf node of a branch, stop searching the current branch and switch to other branches until the traversal is complete.

[0163] 3) New node integration verification: Integrate the new search node into the island, and perform power flow verification and target value calculation through the cooperative control constraint model in combination with the island partitioning objective function; if the model has a solution and the island partitioning objective value meets the requirements, the new node and its secondary loads on the power supply path are successfully integrated; if there is no solution or the target value does not meet the requirements, the current expansion is terminated.

[0164] 4) Iterative loop: Using the updated island as the new root node, return to step 2 and repeat the search until there are no more expansion nodes that meet the conditions.

[0165] 5) Time period connection: Record the current energy storage charge status and EV cluster power energy as the initial state for the next time period, and repeat the above process to realize the dynamic island division by time period.

[0166] Furthermore, to address the issues of slow convergence and susceptibility to local optima in the original algorithm, this invention employs an improved binary antlion algorithm. This algorithm, by introducing interference factors and chaotic mapping, can effectively enhance global search capabilities and convergence speed, enabling rapid and accurate optimization of fault recovery solutions.

[0167] The basic framework of the binary antlion algorithm: The binary antlion algorithm restricts the values ​​of each dimension of the individual ant and antlion to 0 or 1, and updates the ant's position through crossover operations.

[0168]

[0169]

[0170]

[0171] in, An ant individual from generation t+1; , The solution is a sub-solution for random walks; , Solve the d-th dimension data for the child generation; A random number between 0 and 1; , The data for the d-th dimension are ant and parent antlion, respectively. The selection factor decreases with the number of iterations t; This represents the maximum number of iterations.

[0172] Based on the binary antlion algorithm, an interference factor q and chaotic mapping are introduced.

[0173] The interference factor influences the ant's random movement, causing the movement range to gradually decrease from the initial stage to the later stage, thus achieving a balance between local and global search. The interference factor q is defined as follows:

[0174] in, , These are the current and maximum iteration counts, respectively. , The best results are achieved when the time is right.

[0175] The roulette wheel mechanism provides some variability to the algorithm, but it also increases the probability of poorly fit antlions being selected. The chaotic mapping is as follows:

[0176] in, It is a random value between 0 and 1, which will be combined with the chaotic map to complete the chaotic search.

[0177] The specific steps for solving the improved binary antlion algorithm are as follows: (1) Input data: Distribution network topology after island division, photovoltaic output, load data, EV cluster parameters, energy storage parameters, etc.; (2) Initialization parameters: Set the antlion dimension, population size, maximum number of iterations, and weight coefficients for each objective; (3) Population initialization: Construct a binary encoded population to represent decision variables such as switch state, node affiliation, and EV charging / discharging state; (4) Iterative optimization: Calculate the multi-objective optimization function value of each individual, sort the antlion population by fitness, and select elite antlions; guide the ants to walk randomly by combining chaotic mapping and roulette wheel with the parent antlions. (5) Population renewal: Simulate the antlion predation process to update the antlion population and the elite antlion set; (6) Termination judgment: When the maximum number of iterations is reached, the optimal fault recovery scheme is output (including switch operation plan, VSC control mode configuration, photovoltaic-energy storage-EV coordinated control parameters, and islanding division range).

[0178] The standardized recovery process is as follows: (1) Follow the standardized process of "fault clearing - de-idling - restarting collaboration - main network connection" to ensure the security, orderliness and efficiency of island recovery: (2) Fault clearing: Based on the fault detection results, isolate the faulty line, disconnect the faulty branch switch, and determine the scope of the power loss area; (3) De-ionization: Wait for the arc at the fault point to extinguish, ensure the line insulation is restored, and avoid reclosing failure; (4) Restarting the collaboration: Based on the recovery scheme obtained by algorithm optimization, close the interconnection switch, start the photovoltaic-energy storage-EV cluster collaborative control, restore power supply according to load weight priority and island division of targets, and switch VSC control mode synchronously; (5) Main grid connection: After the main grid fault is recovered, the inverter output voltage is synchronized with the main grid voltage in terms of amplitude and phase through VSC pre-synchronization control, and then VSC is smoothly switched to P Q control mode ensures that the grid-connected inrush current is ≤0.3 times the rated current, achieving seamless connection between the island and the main grid and realizing full system recovery.

[0179] This invention focuses on multi-objective optimization, taking into account power supply reliability, operational economy, and EV user satisfaction. By introducing an improved binary antlion algorithm with interference factors and chaotic mapping, it quickly solves the optimal recovery scheme, effectively addressing the shortcomings of traditional algorithms such as slow convergence and susceptibility to local optima. During the recovery process, it fully leverages the synergistic control capabilities of photovoltaic, energy storage, and EV clusters, prioritizing power supply to critical loads and maximizing the utilization of photovoltaic power, achieving a balance of multi-dimensional benefits. Its overall performance is far superior to traditional single-objective recovery strategies.

[0180] Example 2 This embodiment provides a distribution network islanding fault identification and multi-target recovery system, including: The islanding fault state model construction module constructs an islanding fault state model for a photovoltaic distribution network with extremely high penetration rate based on the photovoltaic output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model. The islanding identification and confirmation module, based on the islanding fault state model, uses a hybrid anti-islanding detection method that integrates active disturbance injection and passive feature monitoring to identify and confirm the islanding operation status. According to the islanding operation status, it employs grid-connected inverter power command constraints and VSC multi-mode smooth switching technology to protect the islanded system and perform bidirectional smooth switching between islanded and grid-connected operation. Specifically, the grid-connected inverter power command constraint adjusts the active power command value by calculating the active power command constraint value and combining it with staged dynamic constraint logic. A photovoltaic-storage-EV collaborative constraint construction module is used to determine the photovoltaic-storage-EV collaborative constraint system, which includes island power balance constraints, distribution network island division topology constraints, EV cluster operation constraints, photovoltaic power supply and energy storage operation constraints, and power flow and system security constraints. The multi-objective islanding recovery module, based on the islanding operation status and the photovoltaic-storage-EV collaborative constraint system, determines the optimization objectives as minimizing the power loss load weight, minimizing the number of switching actions, and maximizing EV user satisfaction. It uses dynamic islanding search and an improved binary antlion algorithm to solve the problem and obtain the multi-objective islanding recovery strategy.

[0181] Example 3 This embodiment also provides an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, the processor is connected to the memory, and the one or more computer programs are stored in the memory. When the electronic device is running, the processor executes the one or more computer programs stored in the memory to cause the electronic device to perform the method described in Embodiment 1.

[0182] Example 4 This embodiment also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the method described in Embodiment 1.

[0183] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying islanding faults and restoring multiple targets in a distribution network, characterized in that, include: Based on the photovoltaic power output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model, an islanding fault state model for a photovoltaic distribution network with extremely high penetration rate is constructed. Based on the islanding fault state model, a hybrid anti-islanding detection method is used to determine the operating state of the distribution network, identify and confirm the islanding operating state; Based on the confirmed islanded operation status, a distributed power source and energy storage collaborative control model is constructed; this model includes: a photovoltaic-energy storage-EV collaborative constraint system. Based on the islanded operation status and the photovoltaic-storage-EV collaborative constraint system, the optimization objectives are determined to be minimizing the power loss load weight, minimizing the number of switching actions, and maximizing EV user satisfaction. The multi-objective islanding recovery strategy is obtained by using dynamic islanding search and improved binary antlion algorithm.

2. The method for islanding fault identification and multi-target recovery in a distribution network according to claim 1, characterized in that, Based on the islanded operation status, grid-connected inverter power command constraints and VSC multi-mode smooth switching technology are adopted to protect the islanded system and realize bidirectional smooth switching between islanded operation and grid-connected operation.

3. The method for islanding fault identification and multi-target recovery in a distribution network according to claim 2, characterized in that, The power command constraint of the grid-type inverter is achieved by calculating the active power command constraint value and adjusting the active power command value in combination with the staged dynamic constraint logic.

4. The method for islanding fault identification and multi-target recovery in a distribution network according to claim 1, characterized in that, The photovoltaic-storage-EV collaborative constraint system includes islanded power balance constraints, distribution network islanding topology constraints, EV cluster operation constraints, photovoltaic power supply and energy storage operation constraints, and power flow and system security constraints.

5. The method for islanding fault identification and multi-target recovery in a distribution network according to claim 1, characterized in that, The hybrid anti-islanding detection method is a two-stage collaborative strategy of active disturbance injection and passive feature monitoring.

6. The method for islanding fault identification and multi-target recovery in a distribution network according to claim 5, characterized in that, The dual-stage collaborative strategy of active disturbance injection and passive feature monitoring is as follows: a small frequency disturbance is applied to the output power of the grid-type inverter through active disturbance injection to obtain the detection result; the island voltage amplitude, harmonic distortion rate and power fluctuation characteristics are collected simultaneously through passive feature monitoring; when the voltage amplitude deviation exceeds the first preset threshold, the harmonic distortion rate exceeds the second preset threshold or the power fluctuation amplitude exceeds the third preset threshold, cross-validation is performed with the detection result of active disturbance injection to confirm the islanding status.

7. The method for islanding fault identification and multi-target recovery in a distribution network according to claim 1, characterized in that, Based on the EV individual emergency charging and discharging model and the traffic flow-electricity flow coupling model, an EV cluster aggregated charging and discharging model is constructed.

8. A distribution network islanding fault identification and multi-target recovery system, characterized in that, include: The islanding fault state model construction module constructs an islanding fault state model for a photovoltaic distribution network with extremely high penetration rate based on the photovoltaic output model, weighted load model, EV single-unit emergency charging and discharging model, traffic flow-power flow coupling model, EV cluster aggregated charging and discharging model, and GFM inverter grid-connected system transient model. The islanding identification and confirmation module, based on the islanding fault state model, uses a hybrid anti-islanding detection method to judge the operating status of the distribution network, identify and confirm the islanding operating status; The collaborative control model construction module constructs a collaborative control model for distributed power sources and energy storage based on the confirmed islanded operation status; this model includes: a photovoltaic-energy storage-EV collaborative constraint system; The multi-objective islanding recovery module, based on the islanding operation status and the photovoltaic-storage-EV collaborative constraint system, determines the optimization objectives as minimizing the power loss load weight, minimizing the number of switching actions, and maximizing EV user satisfaction. It uses dynamic islanding search and an improved binary antlion algorithm to solve the problem and obtain the multi-objective islanding recovery strategy.

9. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the processor executes the computer instructions, it performs the steps in the method for identifying and restoring islanded faults in a power distribution network as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, complete the steps in the method for identifying and restoring islanded faults in a power distribution network as described in any one of claims 1-7.

Images