Microgrid resilience enhancement method based on static and dynamic islanding constraints

Abstract

The application discloses a method for enhancing the resilience of a micro-grid under static and dynamic island constraints, comprising: construction of an operation scheduling model in a grid-connected mode; construction of an operation scheduling model in an island mode; and construction of a frequency safety constraint. The island condition triggered by a fault is enhanced to ensure the survival ability of the MG after the island. A nonlinear frequency response model is used, and relevant constraints are combined in a multi-level mixed integer linear model of a planning problem to consider the dynamic frequency behavior after the island.

Description

Technical Field

[0001] This invention relates to a method for resilient enhancement of microgrids based on static and dynamic islanding constraints. Background Technology

[0002] Resilience refers to a system's ability to withstand, adapt to, and rapidly recover from disturbances. With the increasing frequency of natural disasters globally, the construction of "resilient power grids" capable of withstanding extreme disturbances is receiving growing attention. Microgrids are typically characterized by reduced inertia, leading to large transients after unexpected islanding events. These transients can cause cascading devices to disconnect, triggering protection systems and resulting in the complete loss of some loads within the microgrid. Summary of the Invention

[0003] The purpose of this invention is to provide a resilient enhancement method for microgrids based on static and dynamic islanding constraints. This invention proposes a microgrid operation scheduling model for grid-connected operation, which enhances the microgrid through fault-triggered islanding conditions to ensure the survivability (transient and steady-state) of the microgrid after islanding. We use a nonlinear frequency response model and incorporate relevant constraints into a multi-level mixed-integer linear model of the planning problem to consider the dynamic frequency behavior after islanding. Specifically, we include constraints on the maximum rate of frequency change, the minimum frequency point, and the steady-state frequency deviation. Furthermore, to solve this operation planning problem, we propose an iterative solution algorithm to ensure reliable frequency response, self-sufficiency, and optimal operation.

[0004] In the following steps, N, N br T represents the number of nodes, the number of branches, and the planning scope, respectively, and t represents a specific time period. The index {i = 1, ..., N} consists of all nodes in the microgrid, a subset. and These are used for nodes with SG and CIG respectively. Branches are contained within links {ij = 1, ..., N}. br} represents a line from node i to node j, where each link (ij) describes a line from node i to node j. Active and reactive power generated and consumed are represented by p and q, respectively. The superscripts “dg”, “pv”, and “d” represent the power of SGs, CIGs, and loads, respectively. Constant loads and flexible loads are represented by the superscripts “c” and “f” of their respective powers. In grid-connected mode, all predetermined loads must be satisfied, while load shedding is only permitted in islanded mode. The power exchanged with the grid at the point of common coupling is represented as... and System variables include the voltage v at node i. i The active / reactive power flow P between nodes i and j ij / Q ij ; and the net power injection p at node i i / qi Finally, r ij / x ij This represents the resistance / reactance of link ij.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: The present invention proposes a microgrid elastic enhancement method based on static and dynamic islanding constraints, characterized by the following parts:

[0006] Part 1: Construction of the Operation and Scheduling Model under Grid-Connected Mode

[0007] The grid-connected operation model is shown below. The first and second terms of the objective function (a) include the terms added to the active power (C) respectively. grid,p ) and reactive power (C grid,q The cost of power is exchanged with the main grid. Import and export costs vary depending on the energy market. Items three and four are related to SGs (C). pdg C qdg The operating costs of ) are related to the negligible start-up / shutdown costs, while the fifth item (C) is related to the operating costs of ) and the fifth item (C) pv This refers to the operating costs of running and maintaining renewable energy sources. Finally, C... flex It is the penalty cost incurred when the load shifts from a customer's preferred consumption period.

[0008]

[0009] st

[0010]

[0011]

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] Constraints (b)-(c) are the network power flow equations, while (d)-(e) relate to the net power injection at each node. The total load consumed by each node, constant and flexible, is given by (f). Each branch is subject to a maximum load limit S. ij The constraints, by Modeling. The quadratic constraint uses a piecewise approximation of linearization by constructing a convex polygon. Constraints (g)-(h) are modeled to constrain the linearization loading, where... It is the derivative of the eight segments that constitute the convex polygon. Node voltage limits are enforced by (i) and constraints (j)-(m), ensuring that limits on power exchange from the grid, local generation capacity, and total load are not violated. The committed state of the local generator is determined by... and The active power and grid power constraints are shown in equation (j), initially (at ψ=1) based on operator constraints, but these constraints tighten with subsequent iterations based on the solution to the second-stage problem. SGs have up / down (rui / rdi) ramp constraints as shown in equation (n) and minimum on / off time as shown in equation (o), where the parameters are... and The durations of the "on" and "off" periods of SG are defined respectively. The energy provided is limited within the planning scope by equation (p), while the total elastic load energy consumption within one operating cycle... This is guaranteed by equation (q). The grid-connected operation model ψ for each iteration is a mixed-integer linear programming (MILP) problem, where... As a set of control variables.

[0027] Part Two: Construction of the Operation Scheduling Model in Island Mode;

[0028] The goal in the event of unintended islanding is to ensure the self-sufficiency of the microgrid, especially when supplying power to critical loads. Microgrid islanding must guarantee self-sufficiency for at least one time period after disconnection. To achieve this, a robust model considering possible disconnections within each time period of the planning scope is adopted. This problem is addressed independently for each time period where the microgrid may disconnect from the grid and power exchange to the main grid is set to zero in equations (d)-(e). The following equation replaces the objective function (a) to minimize load shedding:

[0029]

[0030] In this mode, although critical loads are always prioritized, all loads can be slashed. When the load on node i is serviced, the integer α... it It is "1" otherwise "0". Indicates load priority and the cost of reducing load at specific nodes. The amount of reduced flexible load is determined by Δ{p,q}. d,f This indicates that constraints on system operation are similar to having... The grid-connected mode. The problem of islanded operation is described as MILP, and its control variables are defined as follows:

[0031] Part Three: Construction of Frequency Security Constraints

[0032] The survival of a microgrid without triggering protection devices after an emergency islanding event depends on the magnitude of the power step change and the control capability of the microgrid generators. Conversely, the power step change is determined by the power exchange with the main grid during disconnection. Survivability constraints embedded in the operation and scheduling problem are extracted from key aspects of the microgrid's dynamic behavior during islanding. Microgrid generators can be grid-supported, capable of providing voltage and frequency control or grid feeding during transient events, with their active and reactive power output determined solely by supervisory control and treated as a constant PQ injection during the transient. The transient response and steady-state operating point after an islanding event are controlled by grid support units (SGs or CIGs) in conjunction with load dynamics. While conventional approaches neglect load dynamics, the methodology can be extended to include their effects. The frequency response of the SG is controlled by electromechanical dynamics and turbine governor dynamics. In a fast-acting CIG, power frequency droop ensures power sharing and frequency control, while inertial response can be simulated by incorporating virtual synchronous machine control. By combining SG and CIG with droop or VSM control using a combined frequency response model, an analytical expression for the performance index of the control transient frequency response when active power undergoes a step change is derived:

[0033] ω(t)=-Δp / m,

[0034]

[0035]

[0036] The characteristics of dynamic frequency response are the instantaneous rate of change of frequency ω(t), the minimum frequency point, and the maximum frequency point (±ω). max The quasi-steady-state response is affected by the frequency deviation ω. s The impact.

[0037] Due to the high penetration rate of grid support units, the reduced system inertia in microgrids affects frequency performance, resulting in larger maximum and minimum values ​​and RoCoF levels. Effective active power management is necessary to prevent activation of underfrequency / overfrequency protection and RoCoF relays.

[0038] Given control parameters and the rated power of the generating units in a grid-connected mode for a given hour, the safe frequency response problem is formulated and solved using a linear programming (LP) problem, as shown in the equation below, for each iteration ψ and time t. To ensure all indices are met, the grid power exchanged at a given time may need to change. Therefore, equation (a) determines the minimum change in grid power dispatch at each moment to obtain a safe dynamic response. Constraints (b)-(d) enforce that grid-connected power plans comply with operator-defined minimum / maximum point, RoCoF, and QSS frequency limits (denoted by lim):

[0039]

[0040] st

[0041]

[0042]

[0043]

[0044] Non-zero optimal cost value This indicates that the previously determined grid connection operation time violated the measurement limit. The value is used to adjust the maximum / minimum power limit exchanged with the grid during the relevant time period in equation (j). The minimum / maximum limits for increasing / decreasing power from / to the grid are shown in the following formula:

[0045]

[0046] Attached Figure Description

[0047] Figure 1 The power generation of the local microgrid generators connected at nodes R1 and R11 in grid-connected and islanded modes;

[0048] Figure 2 For the total network with nominal load curves, load transfer in grid-connected mode (to improve control flexibility) and load reduction in islanded mode;

[0049] Figure 3 This represents the power change from grid input (-) and grid output (+);

[0050] Figure 4 For flexible load scheduling, because the grid power limits are different in each iteration;

[0051] Figure 5 A schematic diagram illustrating the RoCoF and QSS value solutions for each islanding moment. Detailed Implementation

[0052] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0053] The example uses a modified version of the European configuration CIGRE to analyze the performance of the proposed method. The system consists of four photovoltaic (PV) generators and one SG. Three of the PV generators are grid-connected, while one has fixed-output PQ control. Microgrid generator parameters are given in Table 1, with a system baseline of 500 kVA. 50% of the rated load connected to node R1 is movable, and nodes R15 and R16 connect high-priority critical loads (30% of the total load). Load parameters, load curves, and cable parameters are taken from standard parameters, and typical European PV installation power generation curves are considered within a 24-hour planning range. For the dynamic constraints of Phase 2, the lowest frequency point is used. RoCoFω lim =0.8Hz / s and quasi-steady-state frequency The ENTSO-E threshold.

[0054] The CIGRE test case was used to verify and analyze the method for enhancing microgrid resilience under static and dynamic islanding constraints. The generator parameters are shown in the table below:

[0055] Table 1. Generator Set Parameters

[0056] SG Photovoltaic 1 Photovoltaic 2 Photovoltaic 3 Photovoltaic 4 Node R1 R11 R15 R17 R18 kW% peak load 175 350 235 150 90 Inertia (virtual in CIG), H[pu] 7 7 - - - Damping constant, D[pu] 25 30 - - - Mechanical power gain, K[pu] 1.1 1.1 1.1 1.1 - Droop gain, R[pu] 0.03 - 0.05 0.05 - Turbo power fraction, F[pu] 0.35 - - - -

[0057] The specific steps of this invention include:

[0058] Step 1: Operation scheduling calculation in grid-connected mode

[0059] The grid-connected operation model is shown below. The first and second terms of the objective function (a) include the terms added to the active power (C) respectively. grid,p ) and reactive power (C grid,q The cost of power is exchanged with the main grid. Import and export costs vary depending on the energy market. Items three and four are related to SGs (C). pdg C qdg The operating costs of ) are related to the negligible start-up / shutdown costs, while the fifth item (C) is related to the operating costs of ) and the fifth item (C) pv This refers to the operating costs of running and maintaining renewable energy sources. Finally, C... flexIt is the penalty cost incurred when the load shifts from a customer's preferred consumption period.

[0060]

[0061] st

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078] Constraints (b)-(c) are the network power flow equations, while (d)-(e) relate to the net power injection at each node. The total load consumed by each node, constant and flexible, is given by (f). Each branch is subject to a maximum load limit S. ij The constraints, by Modeling. The quadratic constraint uses a piecewise approximation of linearization by constructing a convex polygon. Constraints (g)-(h) are modeled to constrain the linearization loading, where... It is the derivative of the eight segments that constitute the convex polygon. Node voltage limits are enforced by (i) and constraints (j)-(m), ensuring that limits on power exchange from the grid, local generation capacity, and total load are not violated. The committed state of the local generator is determined by... and The active power and grid power constraints are shown in equation (j), initially (at ψ=1) based on operator constraints, but these constraints tighten with subsequent iterations based on the solution to the second-stage problem. SGs have up / down (rui / rdi) ramp constraints as shown in equation (n) and minimum on / off time as shown in equation (o), where the parameters are... and The durations of the "on" and "off" periods of SG are defined respectively. The energy provided is limited within the planning scope by equation (p), while the total elastic load energy consumption within one operating cycle... This is guaranteed by equation (q). The grid-connected operation model ψ for each iteration is a mixed-integer linear programming (MILP) problem, where... As a set of control variables.

[0079] Step 2: Calculation of runtime scheduling in island mode

[0080] The goal in the event of unintended islanding is to ensure the self-sufficiency of the microgrid, especially when supplying power to critical loads. Microgrid islanding must guarantee self-sufficiency for at least one time period after disconnection. To achieve this, a robust model considering possible disconnections within each time period of the planning scope is adopted. This problem is addressed independently for each time period where the microgrid may disconnect from the grid and power exchange to the main grid is set to zero in equations (d)-(e). The following equation replaces the objective function (a) to minimize load shedding:

[0081]

[0082] In this mode, although critical loads are always prioritized, all loads can be slashed. When the load on node i is serviced, the integer α... it It is "1" otherwise "0". Indicates load priority and the cost of reducing load at specific nodes. The amount of reduced flexible load is determined by Δ{p,q}. d,f This indicates that constraints on system operation are similar to having... The grid-connected mode. The problem of islanded operation is described as MILP, and its control variables are defined as follows:

[0083] Step 3: Consider frequency security constraints

[0084] The survival of a microgrid without triggering protection devices after an emergency islanding event depends on the magnitude of the power step change and the control capability of the microgrid generators. Conversely, the power step change is determined by the power exchange with the main grid during disconnection. Survivability constraints embedded in the operation and scheduling problem are extracted from key aspects of the microgrid's dynamic behavior during islanding. Microgrid generators can be grid-supported, capable of providing voltage and frequency control or grid feeding during transient events, with their active and reactive power output determined solely by supervisory control and treated as a constant PQ injection during the transient. The transient response and steady-state operating point after an islanding event are controlled by grid support units (SGs or CIGs) in conjunction with load dynamics. While conventional approaches neglect load dynamics, the methodology can be extended to include their effects. The frequency response of the SG is controlled by electromechanical dynamics and turbine governor dynamics. In a fast-acting CIG, power frequency droop ensures power sharing and frequency control, while inertial response can be simulated by incorporating virtual synchronous machine control. By combining SG and CIG with droop or VSM control using a combined frequency response model, an analytical expression for the performance index of the control transient frequency response when active power undergoes a step change is derived:

[0085] ω(t)=-Δp / m,

[0086]

[0087]

[0088] The characteristics of dynamic frequency response are the instantaneous rate of change of frequency ω(t), the minimum frequency point, and the maximum frequency point (±ω). max The quasi-steady-state response is affected by the frequency deviation ω. s The impact.

[0089] Due to the high penetration rate of grid support units, the reduced system inertia in microgrids affects frequency performance, resulting in larger maximum and minimum values ​​and RoCoF levels. Effective active power management is necessary to prevent activation of underfrequency / overfrequency protection and RoCoF relays.

[0090] Given control parameters and the rated power of the generating units in a grid-connected mode for a given hour, the safe frequency response problem is formulated and solved using a linear programming (LP) problem, as shown in the equation below, for each iteration ψ and time t. To ensure all indices are met, the grid power exchanged at a given time may need to change. Therefore, equation (a) determines the minimum change in grid power dispatch at each moment to obtain a safe dynamic response. Constraints (b)-(d) enforce that grid-connected power plans comply with operator-defined minimum / maximum point, RoCoF, and QSS frequency limits (denoted by lim):

[0091]

[0092] st

[0093]

[0094]

[0095]

[0096] Non-zero optimal cost value This indicates that the previously determined grid connection operation time violated the measurement limit. The value is used to adjust the maximum / minimum power limit exchanged with the grid during the relevant time period in equation (j). The minimum / maximum limits for increasing / decreasing power from / to the grid are shown in the following formula:

[0097]

[0098]

[0099] Figure 1 The power generation of the local microgrid generators connected at nodes R1 and R11 in both grid-connected and islanded modes is given. The power generation of the local microgrid generators connected at nodes R1 and R11 in both grid-connected and islanded modes can be found in the appendix of the instruction manual. Figure 1 As shown, Figure 1 This paper shows the hourly power output of two generators (one SG and one PV, connected to nodes R1 and R11 respectively) in grid-connected and islanded modes. In grid-connected operation, the goal is to minimize operating costs while meeting load demand. Since the power generation cost of the PV devices is zero, their output is maximized. In islanded mode, the microgrid should have sufficient power generation capacity to serve critical loads. The hourly MG sufficiency is analyzed over 24 hours based on the PV and SG energy content present at a given time. Figure 1 As can be seen, SG is only used during periods of insufficient solar energy. Furthermore, due to excessive photovoltaic power generation during MG islanding, the variability of PV power leads to a reduction in the active power of PV units observed in islanding mode between 9 and 14 hours.

[0100] Figure 2 The overall network with nominal load curves is presented, along with load transfer in grid-connected mode (improving control flexibility) and load reduction in islanded mode. Variable photovoltaic power generation and insufficient SG (storage grid) lead to load reduction during certain periods, such as... Figure 2As shown. This primarily occurs within 20 to 24 hours, which is part of the peak consumption period (18 to 24 hours). However, the load reduction is limited to a maximum of 40% in each case, with the node's critical load primarily serving under emergency islanding conditions. The results demonstrate the adequacy of the MG network, indicating that better power management of PV units is necessary to improve reliability and better support islanding power modes.

[0101] Figure 3 A diagram illustrating the power changes from grid input (-) and grid output (+) is provided. Please note that... Figure 3 As indicated by the positive value of the grid power, in grid-connected mode, any excess photovoltaic power will be sold to the grid. For example... Figure 3 , 4 The diagram shows that the operating cost in grid-connected mode is minimized by reducing the time period during which the load is transferred to the system and excess power is generated from the PV units. This minimizes the microgrid's dependence on grid power and provides greater flexibility, especially considering the limitations imposed by grid power exchange. To minimize performance degradation and prevent cascading failures due to the operation of protective relays during landing, the operating schedule is tested to ensure that violations of dynamic constraints are eliminated. The grid power is initially planned as follows: Figure 3 Iteration 1 is shown in the figure. Iterations 2 and 3 show that the planned grid power is reduced because the system's control capability is insufficient to meet the dynamic constraints.

[0102] Figure 4 Flexible load scheduling is provided, as illustrated by the grid power limits for each iteration. Dynamic frequency control capability is governed by nominal active power capacity and unit control parameters, as defined in Table I. Since these parameters are static, further system flexibility is crucial. Figure 4 The results show that using flexible loads increases system redundancy and prevents microgrid model infeasibility, where insufficient control capabilities cause the frequency response to fail to meet thresholds. These are activated as preventative controls in iterations 2 and 3 to improve survivability during emergency islanding events.

[0103] The frequency values ​​of RoCoF and QSS changes per islanding cycle are as follows: Figure 5 As shown. Iterative solutions for RoCoF and QSS values ​​at each islanding moment. Positive / negative values ​​related to grid active power output / input before microgrid disconnection.

Claims

1. A microgrid resilient enhancement method based on static and dynamic islanding constraints, characterized in that, The method includes: S1, Construction of the operation scheduling model under grid connection mode; S2, the construction of the operation scheduling model in the island mode; S3, Construction of frequency security constraints; S4, Iterative solution to the runtime scheduling problem; S1 includes: Scheduling is performed in grid-connected mode based on preset conditions; The preset conditions include: the amount of active power added to the active power and reactive power exchanged with the main grid, import and export costs and operating costs, the operating costs of operating and maintaining renewable energy, and the penalty costs of shifting loads from the customer's preferred consumption period; The grid-connected operation model is shown below; The first and second terms of objective function a) include the active power exchanged with the main power grid, respectively. and reactive power Import and export costs vary depending on the energy market; items three and four are related to... The fifth item is related to operating costs. It is the operating cost of running and maintaining renewable energy; This is the penalty cost incurred when the load shifts from a customer's preferred consumption period; (a)， st (b)， (c)， (d)， (e)， (f)， (g)， (h)， (i)， (j)， (k)， (l)， (m)， (n)， (o)， (p)， (q)， Constraint bc is the network power flow equation, while de is related to the net power injection of each node; the total load consumed by each node is constant and flexible, given by f; each branch is subject to a maximum load limit. The constraints, by Modeling; the quadratic constraint uses a piecewise approximation of linearization by constructing a convex polygon; constraint gh models the linearization loading limit, where It is the derivative of the octet that constitutes the convex polygon; the node voltage limit is enforced by i and constraint jm, ensuring that the limits on power exchange from the grid, local generation capacity, and total load are not violated; the commitment state of the local generator is determined by... and The active power and grid power constraints are shown in Equation j. Initially, at ψ=1, these constraints are based on operator limitations. However, with subsequent iterations, these constraints are tightened based on the solution to the second-stage problem. SGs has upward / downward rui / rdi ramp constraints as shown in Equation n and minimum on / off time as shown in Equation o, where the parameters are... and The durations of the "on" and "off" periods of SG are defined respectively; The energy provided is limited by formula p within the planning range, while the total elastic load energy consumption within one operating cycle... The grid-connected operation model for each iteration is guaranteed by equation q. This is a mixed-integer linear programming (MILP) problem, where As a set of control variables; Step S2 includes: In the islanded mode scheduling strategy, it is guaranteed that the microgrid will be self-sufficient for at least a certain period of time after disconnection: Microgrid islands must maintain self-sufficiency for at least one time period after disconnection. A robust model considering potential disconnections within each time period of the planning scope is adopted. For each time period where the microgrid may disconnect from the grid and power exchange to the main grid is set to zero in equations d-e, this problem is solved independently. The following equation replaces the objective function a to minimize load shedding: ， In this mode, while critical loads are always prioritized, all loads are slashed; when the load on node i is serviced, integers are used. If the value is "1", then the value is "0"; Indicates load priority and the cost of reducing load at specific nodes; the reduced flexible load is determined by... This indicates that the constraints on system operation are similar to having The grid-connected mode; the problem of islanded operation is described as MILP, and its control variables are defined as .

2. The microgrid resilience enhancement method based on static and dynamic islanding constraints according to claim 1, characterized in that, S3 includes: In fast-acting CIG, power frequency droop ensures power sharing and frequency control, while inertial response is simulated by combining virtual synchronous machine control. By using a combined frequency response model to combine SG and CIG with droop or VSM control, an analytical expression for the performance index of controlling transient frequency response during step changes in active power is derived. ， ， ， The characteristic of dynamic frequency response is the instantaneous rate of change of frequency. The lowest and highest frequency points ± The quasi-steady-state response is affected by frequency deviation. The impact; Due to the high penetration rate of grid support units, the reduced system inertia in microgrids can affect frequency performance, resulting in larger maximum and minimum values ​​and RoCoF levels. To prevent the activation of underfrequency / overfrequency protection and RoCoF relays, effective management of active power is required. Given control parameters and the rated power of the generating units in a grid-connected mode for a given hour; for each iteration ψ and time t, formulate and solve the safe frequency response problem using the linear programming (LP) problem shown below; to ensure that all indices are met, the grid power exchanged at a given time may need to change. Therefore, equation a determines the minimum change in grid power dispatch at each moment to obtain a safe dynamic response; constraints b-d enforce that grid-connected power plans comply with operator-defined minimum / maximum point, RoCoF, and QSS frequency limits. express: ， st ， ， ， Non-zero optimal cost value This indicates that the previously determined grid connection operation time violated the measurement limit; The value is used to adjust the maximum / minimum power limit exchanged with the grid in equation j within the relevant time period; The minimum / maximum limits for increasing / decreasing power from / to the grid are shown in the following formula: ， 。

Images