A rolling optimization method for distribution network fault restoration considering the randomness of load and photovoltaic power

Abstract

A rolling optimization strategy method for distribution network fault restoration considering the randomness of load and photovoltaic power. A random model of photovoltaic load is constructed, where the random prediction error of load is described by Gaussian distribution, and the random prediction error of photovoltaic power output is described by normal distribution. A power flow constraint model, a node ZIP load constraint model, an on-load tap changer (OLTC) regulation model, a topology and virtual power flow constraint model, and a soft open point (SOP) model are constructed. A two-layer optimization and reconstruction model is constructed. The upper layer of the model takes the distribution network architecture, the actions of on-load tap changers, and the switching of capacitor banks as decision variables. The optimization objective of the lower layer is the active and reactive power output of distributed energy sources, the power magnitude of load shedding, and the system power loss during this period. By setting objective functions with the same mathematical form, distribution network fault reconstruction and power flow optimization are achieved. The present invention improves the consumption rate of distributed power sources, reduces system power loss, and improves the power recovery ability of the system under fault conditions.

Description

Technical Field

[0001] The present invention relates to a rolling optimization method for distribution network fault restoration considering the randomness of load and photovoltaic power Background Art

[0002] The fault self-healing of a distribution network means that after the fault location and isolation of the distribution network, by opening and closing its tie switches and sectionalizing switches, the topological structure of the power grid is changed to achieve the rapid restoration of the non-fault area. The existing research on traditional distribution networks has been very rich. In many studies, various methods including sectionalizing switches, remotely controllable switches, and dispatchable distributed generators (DG) have been used to establish an active distribution network fault reconstruction model. However, as a new type of power electronic device, the soft open point (SOP) has gradually replaced the traditional tie switch, bringing great potential to improve the operation flexibility of the distribution network. The SOP has the ability to transfer load, and its fault-side converter can be switched to a virtual power source to re-energize the affected area, thereby further improving the system's restoration ability. However, in many studies, the load and DG output after the fault are usually regarded as fixed values, without considering scenarios such as node voltage violation that may be caused by the random changes in DG output and load. Therefore, its reconstruction scheme often cannot meet the system operation constraints Summary of the Invention

[0003] In order to overcome the above deficiencies of the prior art, the present invention proposes a two-layer rolling optimization strategy for distribution network fault restoration with SOP considering the uncertainty of load and photovoltaic output. The upper layer of the model takes the distribution network architecture, the operation of on-load tap changers (OLTCs), and the switching of capacitor banks as decision variables. The optimization objectives of the lower layer are the active and reactive power outputs of DGs, the power magnitude of load shedding, and the system network loss. By setting objective functions with the same mathematical form (the objective functions are full-time optimization and subsequent-time optimization respectively), the distribution network fault reconstruction and power flow optimization are realized. Secondly, this model essentially belongs to a large-scale linear programming model, and the second-order cone programming method is used for model transformation and solution. Finally, by performing model transformation and solution on an example, the factors affecting the distribution network fault restoration ability are obtained, and the effectiveness of the proposed method is verified

[0004] In order to achieve the above object, the technical solution of the present invention is as follows

[0005] A rolling optimization method for distribution network fault restoration considering the randomness of load and photovoltaic power, characterized in that the method includes the following steps

[0006] S1: Construct a photovoltaic load randomness model. Use Gaussian distribution to describe the random prediction error of the load and normal distribution to describe the random prediction error of the photovoltaic power generation output.

[0007] S2: Construct a power flow constraint model, including a node ZIP load constraint model, an OLTC regulation model, a topology and virtual power flow constraint model, and an intelligent soft switch SOP model.

[0008] S3: Construct a two-layer optimization and reconstruction model. The upper layer of the model takes the distribution network architecture, the actions of the OLTC, and the switching of capacitor banks as decision variables. The optimization objective of the lower layer is the active and reactive power output of distributed energy sources, the power magnitude of load shedding, and the system power loss during this period. By setting objective functions with the same mathematical form (the objective functions are for full-period optimization and subsequent-period optimization respectively), achieve distribution network fault reconstruction and power flow optimization.

[0009] S4: Obtain the relevant parameters required for optimization and solve the model.

[0010] Furthermore, in step S1, the modeling of load randomness and photovoltaic power generation randomness specifically includes:

[0011] S1-1: The daily load data of the distribution network is obtained from load forecasting. Use Gaussian distribution to describe the random prediction error of the load. Then the probability density function of the load data is:

[0012]

[0013]

[0014] where P LD and Q LD are the active and reactive power outputs of the load, μ LD is the output prediction value, and σ LD is the standard deviation of the prediction error. Generate 500 groups of random output scenarios through this probability density function and reduce them to 20 groups of scenarios for subsequent case studies through backward reduction.

[0015] S1-2: The photovoltaic output can be predicted according to meteorological conditions. Use normal distribution to describe the random prediction error of the photovoltaic power generation output. Then the probability density function of the active power output of the photovoltaic power generation is:

[0016]

[0017] where P PV is the active power of the photovoltaic output, μ PV is the output prediction value, and σ PV is the standard deviation of the prediction error. Generate 500 groups of random output scenarios through this probability density function and reduce them to 20 groups of scenarios for subsequent case studies through backward reduction.

[0018] Furthermore, in the step S2, a power flow constraint model, a node ZIP load constraint model, an OLTC regulation model, a topology and virtual power flow constraint model, and an SOP model are constructed, specifically including:

[0019] S2-1: The distribution network power flow model is a power flow equation established based on branch power. Compared with the traditional power flow calculation based on node power, the power flow model is more suitable for power flow calculation of radial distribution systems. In the distribution network reconfiguration problem, due to the problem of changes in branch opening and closing conditions, it is necessary to improve the traditional power flow model by introducing the line opening variable Z ij Relax the power flow equation and select the π-type equivalent line model to obtain the following power flow equation suitable for distribution network reconfiguration:

[0020]

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] where Ω s [[ID=3']]is all the conducting branches, r ij and x ij are the resistance and reactance of branch ij, is the admittance of branch ij, Z ij is the opening variable of branch ij, Z ij = 1 represents that branch ij is in the open state, which is the prerequisite for the realization of formula (5) and formula (6); P ij,ω,t is the active power flowing from node i to node j on the branch; Q ij,ω,t is the reactive power flowing from node i to node j on the branch; is the square value of the current flowing from node i to node j on the branch; is the square value of the voltage of node i; and are the active and reactive power of the node when power is lost. is the active power output at the SOP port at node i, is the reactive power output at the SOP port at node i. and are the active power injected by distributed power sources and consumed by loads at node i respectively; is the photovoltaic power generation at the node, is the photovoltaic power waste at the node. Q i,ω,t is the sum of the reactive power injected at node i, P i,ω,t is the sum of the active power injected at node i. and are the reactive power injected by the distributed power source and consumed by the load at node i respectively. is the reactive power provided by the capacitor bank; Constraints on the upper and lower limits of the active and reactive power outputs of various devices:

[0027] Line power constraints:

[0028]

[0029]

[0030] Node voltage and current constraints:

[0031]

[0032]

[0033] Power loss constraints:

[0034]

[0035] Generator constraints:

[0036]

[0037]

[0038] Wind and photovoltaic power waste constraints:

[0039]

[0040] Among them, P ji , respectively represent the minimum and maximum active power of the line, Q ji , respectively represent the minimum and maximum reactive power of the line. U i , respectively represent the minimum and maximum voltages of the node. respectively represent the maximum active power and reactive power of the generator;

[0041] S2-2: The most widely accepted load model for the distribution network at the present stage is the ZIP model, which divides the load power demand into three parts: constant impedance (Z), constant current (I), and constant power (P). Then the load can be expressed in the following form:

[0042]

[0043]

[0044] where k p,1 +k p,2 +k p,3 =1, k q,1 +k q,2 +k q,3 =1, is the rated active power value at the rated voltage U N ; is the rated power value at the rated voltage U N ;

[0045] It can be seen that the nodal power is a non - linear function of the nodal voltage and cannot be incorporated into the linear programming model used in this paper. CVR is defined as the ratio of the percentage of active or reactive power to the percentage reduction in bus voltage. From the definition of CVR, an equivalent model for voltage - sensitive loads can be obtained, and the formula is as follows:

[0046]

[0047]

[0048] where and also Considering U i,ω,t ≈U N and it can be obtained that:

[0049]

[0050]

[0051] At this time, equations (17) and (18) have been linearized and can be included in the power flow constraints, and the value of CVR can be estimated from the ZIP coefficients of the load. Examining the original model and setting U N =1 p.u., it can be obtained that:

[0052]

[0053]

[0054] Through equations (23), (24) and the assumption U i,ω,t ≈U N , the value of CVR can be obtained as follows:

[0055] CVR P =2k P,1 +k P,2 (25)

[0056] CVR Q = 2k Q,1 + k Q,2 (26)

[0057] S2-3: The OLTC can adjust the output voltage within a certain range by adjusting the tap position. The model is established as follows:

[0058]

[0059]

[0060]

[0061] Among them, represents the tap position at time t, is the square value of the voltage corresponding to the tap position, is the square value of the OLTC output voltage, represents the square of the per-unit value of the voltage at node 1 at time t, U respectively represent the upper and lower limits of the node voltage at that place;

[0062] S2-4: The reconfiguration of the distribution network needs to ensure the connectivity of the reconfigured distribution system, and there are no islands and loops. Therefore, the connectivity and radiality constraints of the distribution system can be expressed as:

[0063]

[0064] X 12,t = z 12,t (31)

[0065]

[0066]

[0067]

[0068] Among them, E is the set of line nodes, z ij,t is the line outage variable, X ij,t is the parent-child node representation variable, X ij,t = 1 indicates that i is the parent node of j at time t, λ j,t is the virtual power flow demand of the node, is the virtual power flow supply of the generator, is the virtual line power, c(j), δ(j) represent the adjacent nodes of j, Ω DG represents the set of nodes connected to the distributed generator, Ω SOP represents the set of nodes connected to the SOP, ΩG It represents the set of nodes connected to the superior power grid. Formulas (32)-(34) indicate that the parent node of a node is at most one and must be greater than its own virtual power flow demand;

[0069] S2-5: SOP is a new type of intelligent distribution device that replaces the traditional tie switch. Its application will greatly improve the flexibility and controllability of the operation of the distribution system. However, the role of SOP in the self-healing process of the distribution network fault has been less studied. Compared with the tie switch, the power control of SOP is more accurate and reliable, avoiding the potential safety hazards that may be brought by switch operations. When a fault occurs, due to the role of DC isolation, it can effectively prevent the fault current from crossing; during the power supply restoration process, it can provide effective voltage support for the power-loss side, thus expanding the power supply restoration range. At present, the back-to-back voltage source converter (B2B VSC) type SOP is a relatively commonly used SOP, and its topological structure is realized by connecting two converters through a DC capacitor.

[0070] Adding SOP to the branches of the distribution system can, to a certain extent, improve the power flow distribution of the distribution system, reduce the system network loss, and balance the network voltage. After a fault occurs in the distribution network, a power-loss area is formed after fault location and isolation. As a distribution device that replaces the tie switch, SOP cannot restore power supply to any power-loss area in the entire distribution system. The following analyzes the power-loss areas where SOP can be used for power supply restoration.

[0071] After a fault in the distribution network, the power-loss areas formed after fault location and isolation can be divided into the following 3 cases according to the SOP access position:

[0072] 1) Neither end of the SOP is in the power-loss area;

[0073] 2) One end of the SOP is in the power-loss area and the other end is not in the power-loss area;

[0074] 3) Both ends of the SOP are in the power-loss area.

[0075] When a certain area can be connected to the superior network through the tie switch, the SOP operates in the PQ control mode and optimizes its active power output and reactive power output. When the power-loss area cannot be connected to the superior network through the tie switch, the SOP adopts V f control mode. At this time, the outlet voltage of the distribution network SOP and the switch states in the distribution network should be adjusted for fault recovery, and the remaining unrecovered loads form corresponding islands. Considering scenarios 1 and 2 comprehensively, the following SOP constraints are proposed:

[0076]

[0077]

[0078]

[0079] SOP power-off side voltage constraint:

[0080]

[0081] Among them, is the loss coefficient of SOP; is the SOP loss connected to node i; is the SOP capacity connected to node i; Ω n represents the set of power-off side nodes; U0 is the minimum limit value of the per-unit value of the power-off side node voltage, generally taken as 1.0; Equation (35) is the SOP active power constraint, Equation (36) is the SOP loss constraint, Equation (37) is the SOP capacity constraint, and Equation (38) indicates that if one end of the SOP is located on the power-off side, it is adjusted to V f Control mode;

[0082] In the step S3, the double-layer optimization and reconstruction model specifically includes:

[0083] S3-1: The upper-layer objective function is in the following form:

[0084]

[0085] Among them, α is the power-off cost coefficient, β is the wind and light abandonment cost coefficient, and γ is the SOP loss cost coefficient;

[0086] The goal of the upper-layer model is to complete the planning of the distribution network switch operation, OLTC operation, and capacitor bank switching every 1 hour at the moment and shortly after the fault occurs, considering the randomness of the load and photovoltaic power generation. S3-2: The lower-layer objective function is in the following form:

[0087]

[0088] The goal of the lower-layer model is to obtain the optimal solutions of the DG output, load shedding size, and network and SOP loss conditions within the next 15 minutes based on the current load and photovoltaic power generation forecasts, on the basis of determining the above equipment actions, and to maximize the power supply capacity of the power grid and reduce the system loss.

[0089] The objective functions described in the step S3 are full-time optimization and subsequent-time optimization respectively.

[0090] The present invention constructs a photovoltaic load randomness model, uses Gaussian distribution to describe the random prediction error of the load, and normal distribution to describe the random prediction error of the photovoltaic power generation output; constructs a power flow constraint model, a node ZIP load constraint model, an on-load tap changer (OLTC) regulation model, a topology and virtual power flow constraint model, and a soft open point (SOP) model; constructs a two-layer optimization and reconstruction model. The upper layer of the model takes the distribution network architecture, the operation of the on-load tap changer, and the switching of capacitor banks as decision variables, and the optimization objective of the lower layer is the active and reactive power output of distributed energy sources at this time period, the power magnitude of load shedding, and the system power loss magnitude. By setting objective functions with the same mathematical form (the objective functions are full-time period optimization and subsequent time period optimization respectively), the distribution network fault reconstruction and power flow optimization are realized. The present invention considers that the distribution network after network reconstruction is beneficial to reducing the phenomenon of abandoned electricity of distributed power sources and improving the accommodation rate of distributed power sources. At the same time, it also considers the division and effect of the two-stage decision-making work of distribution network reconstruction before and within the day. The power grid reconstruction plan can be obtained before the day, greatly reducing the calculation time of the decision-making within the day and taking into account the calculation accuracy. In addition, by making full use of the control of the power flow by SOP, the system power loss is reduced, and the power recovery ability of the system under fault conditions is improved.

[0091] The beneficial effects of the present invention are as follows:

[0092] 1) Considering the distribution network after network reconstruction is beneficial to reducing the phenomenon of abandoned electricity of distributed power sources and improving the accommodation rate of distributed power sources.

[0093] 2) Considering the division and effect of the two-stage decision-making work of distribution network reconstruction before and within the day, the power grid reconstruction plan can be obtained before the day, greatly reducing the calculation time of the decision-making within the day and taking into account the calculation accuracy.

[0094] 3) Considering the influence of SOP on distribution network reconstruction, by controlling the power flow by SOP, the system power loss is reduced, and the power recovery situation of the system under fault conditions is improved. Description of the Drawings

[0095] Figure 1 is a schematic diagram of a typical distribution system fault scenario.

[0096] Figure 2 is a schematic diagram of the example simulation of the present invention.

[0097] Figure 3 is a schematic diagram of the total output of new energy equipment of the present invention.

[0098] Figure 4 is a system voltage distribution diagram at the 17th moment of the present invention.

[0099] Figure 5 It is the loss diagram of each period of the system of the present invention.

[0100] Figure 6 It is the system voltage diagram of the present invention.

[0101] Figure 7 It is the statistical chart of the second-order cone conversion error of the present invention.

[0102] Figure 8 It is the flow chart of the method of the present invention. Specific implementation method

[0103] The present invention will be further described below in conjunction with the accompanying drawings.

[0104] Refer to Figures 1 to 8 , a rolling optimization strategy method for distribution network fault restoration considering the randomness of load and photovoltaic power generation, includes the following steps:

[0105] S1: Construct a randomness model of photovoltaic load, use Gaussian distribution to describe the random prediction error of load, and normal distribution to describe the random prediction error of photovoltaic power generation output;

[0106] S2: Construct a power flow constraint model, a node ZIP load constraint model, an OLTC regulation model, a topology and virtual power flow constraint model, and an SOP model;

[0107] S3: Construct a two-layer optimization reconstruction model. The upper layer of the model uses the distribution network architecture, the operation of on-load tap-changing transformers, and the switching of capacitor banks as decision variables. The optimization goal of the lower layer is the active and reactive power output of distributed energy sources, the power magnitude of load shedding, and the system power loss magnitude at this time period. By setting objective functions with the same mathematical form (the objective functions are full-time period optimization and subsequent time period optimization respectively), distribution network fault reconstruction and power flow optimization are realized.;

[0108] S4: Obtain the relevant parameters required for optimization and solve the model.

[0109] Furthermore, in the step S1, the modeling of load randomness and photovoltaic power generation randomness includes the following components:

[0110] S1-1: The daily load data of the distribution network is obtained from load forecasting. Gaussian distribution is used to describe the random prediction error of load, and the probability density function of load data is:

[0111]

[0112]

[0113] Where P LD 、Q LD are the active and reactive power outputs of the load, and μ LDis the predicted output value, and σ LD is the standard deviation of the prediction error. 500 sets of random output scenarios are generated through this probability density function and reduced to 20 sets of scenarios for subsequent examples through backward reduction;

[0114] S1-2: The photovoltaic output can be predicted according to meteorological conditions. The random prediction error of photovoltaic power generation output is described by a normal distribution. Then the probability density function of the active power output of photovoltaic power generation is:

[0115]

[0116] where P PV is the active power of the photovoltaic output, μ PV is the predicted output value, and σ PV is the standard deviation of the prediction error. 500 sets of random output scenarios are generated through this probability density function and reduced to 20 sets of scenarios for subsequent examples through backward reduction;

[0117] Furthermore, in the step S2, a power flow constraint model, a node ZIP load constraint model, an OLTC regulation model, a topology and virtual power flow constraint model, and an SOP model are constructed:

[0118] S2-1: The power flow model of the distribution network is a power flow equation established based on branch power. Compared with the traditional power flow calculation based on node power, the power flow model is more suitable for the power flow calculation of radial distribution systems. In the problem of distribution network reconfiguration, due to the problem of changes in branch opening and closing conditions, it is necessary to improve the traditional power flow model, introduce the line opening variable Z ij relax the power flow equation, and select the π-type equivalent line model to obtain the following power flow equation suitable for distribution network reconfiguration:

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125] where, Ω s is all the conducting branches, r ij and x ij are the resistance and reactance of branch ij, is the admittance of branch ij, and Z ijis the switching variable of branch ij, Z ij = 1 represents that branch ij is in the open state, which is the prerequisite for the implementation of formulas (5) and (6); P ij,ω,t is the active power flowing from node i to node j on the branch; Q ij,ω,t is the reactive power flowing from node i to node j on the branch; is the square value of the current flowing from node i to node j on the branch; is the square value of the voltage at node i; and are the active and reactive power of the node when power is lost. is the active power output of the SOP port at node i, is the reactive power output of the SOP port at node i. and are the active power injected by distributed power sources and consumed by loads at node i respectively; is the photovoltaic power generation at the node, is the photovoltaic curtailment power at the node. Q i,ω,t is the sum of the reactive power injected at node i, P i,ω,t is the sum of the active power injected at node i. and are the reactive power injected by distributed power sources and consumed by loads at node i respectively. is the reactive power provided by the capacitor bank;

[0126] Constraints on the upper and lower limits of the active and reactive power outputs of various devices:

[0127] Line power constraints:

[0128]

[0129]

[0130] Node voltage and current constraints:

[0131]

[0132]

[0133] Power loss constraints:

[0134]

[0135] Generator constraints:

[0136]

[0137]

[0138] Wind and photovoltaic curtailment constraints:

[0139]

[0140] Among them, P ji , respectively represent the minimum and maximum active power of the line, Q ji , respectively represent the minimum and maximum reactive power of the line. U i , respectively represent the minimum and maximum voltages of the node. respectively represent the maximum active and reactive power of the generator;

[0141] S2-2: The most widely accepted load model for the distribution network at the present stage is the ZIP model. It divides the load power demand into three parts: constant impedance (Z), constant current (I), and constant power (P). Then the load can be expressed in the following form:

[0142]

[0143]

[0144] Among them, k p,1 +k p,2 +k p,3 =1, k q,1 +k q,2 +k q,3 =1, is the rated active power value at the rated voltage U N , is the rated power value at the rated voltage U N .

[0145] It can be seen that the node power is a non-linear function of the node voltage and cannot be incorporated into the linear programming model used in this paper. CVR is defined as the ratio of the percentage of active or reactive power to the percentage reduction of the bus voltage. From the definition of CVR, an equivalent model for voltage-sensitive loads can be obtained, and the formula is as follows:

[0146]

[0147]

[0148] Among them, Also,

[0149] Considering U i,ω,t ≈U N and it can be obtained that:

[0150]

[0151]

[0152] At this time, equations (17) and (18) have been linearized and can be included in the power flow constraints, and the value of CVR can be estimated from the ZIP coefficients of the load. Examine the original model and set U N = 1 p.u., the following can be obtained:

[0153]

[0154]

[0155] Through equations (23), (24) and the assumption that U i,ω,t ≈U N , the value of CVR can be obtained as follows:

[0156] CVR P = 2k P,1 + k P,2 (25)

[0157] CVR Q = 2k Q,1 + k Q,2 (26)

[0158] S2-3: The OLTC can adjust the output voltage within a certain range by adjusting the tap position. The model is established as follows:

[0159]

[0160]

[0161]

[0162] where, represents the tap position at time t, is the square value of the voltage corresponding to the tap position, is the square value of the output voltage of the OLTC, represents the square of the per-unit value of the voltage at node 1 at time t, U respectively represent the upper and lower limits of the node voltage at that place;

[0163] S2-4:: The distribution network reconstruction needs to ensure the connectivity of the reconstructed distribution system, and there are no islands and loops. Therefore, the distribution system connectivity and radiation constraints can be expressed as:

[0164]

[0165] X 12,t = z12,t (31)

[0166]

[0167]

[0168]

[0169] Among them, E is the set of line nodes, z ij,t is the line outage variable, X ij,t is the parent-child node representation variable, X ij,t = 1 indicates that at time t, i is the parent node of j, λ j,t is the virtual power flow demand of the node, is the virtual power flow supply of the generator, is the virtual line power, c(j) and δ(j) represent the adjacent nodes of j, Ω DG represents the set of nodes connected to the distributed generator, Ω SOP represents the set of nodes connected to the SOP, Ω G represents the set of nodes connected to the superior power grid. Formulas (32)-(34) indicate that a node has at most one parent node and must be greater than its own virtual power flow demand;

[0170] S2-5: The SOP is a new type of intelligent distribution device that replaces the traditional tie switch. Its application will greatly improve the flexibility and controllability of the distribution system operation. However, the role of the SOP in the self-healing process of the distribution network has been less studied. Compared with the tie switch, the power control of the SOP is more accurate and reliable, avoiding potential safety hazards that may be brought by switch operations. When a fault occurs, due to the effect of DC isolation, it can effectively prevent the fault current from crossing; during the power supply restoration process, it can provide effective voltage support for the power-loss side, thereby expanding the power supply restoration range. Currently, the back-to-back voltage source converter (B2B VSC)-type SOP is a commonly used SOP, and its topological structure is realized by connecting two converters through a DC capacitor.

[0171] Adding the SOP to the branches of the distribution system can, to a certain extent, improve the power flow distribution of the distribution system, reduce the system network loss, and balance the network voltage. After a fault occurs in the distribution network, a power-loss area is formed after fault location and isolation. As a distribution device that replaces the tie switch, the SOP cannot restore power supply to any power-loss area in the entire distribution system. The following analyzes the power-loss areas where the SOP can be used for power supply restoration.

[0172] Such as Figure 1As shown in the figure, after a distribution network fault, the power outage area formed by fault location and fault isolation can be divided into the following three situations according to the SOP access position:

[0173] 1) Both ends of the SOP are not in the power outage area;

[0174] 2) One end of the SOP is in the power outage area and the other end is not in the power outage area;

[0175] 3) Both ends of the SOP are in the power outage area.

[0176] When a certain area can be connected to the upper-level network through the tie switch, the SOP operates in the PQ control mode and optimizes its active power output and reactive power output. When the power outage area cannot be connected to the upper-level network through the tie switch, the SOP adopts V f control mode. At this time, the outlet voltage of the distribution network SOP and the switch state in the distribution network should be adjusted for fault recovery, and the remaining unrecovered loads form corresponding islands. Considering scenarios 1 and 2 comprehensively, the following SOP constraints are proposed:

[0177]

[0178]

[0179]

[0180] SOP power outage side voltage constraint:

[0181]

[0182] Among them, is the loss coefficient of the SOP; is the SOP loss connected to node i; is the SOP capacity connected to node i; Ω n represents the set of nodes on the power outage side; U0 is the minimum limit value of the per-unit value of the node voltage on the power outage side, generally taken as 1.0; Equation (35) is the SOP active power constraint, Equation (36) is the SOP loss constraint, Equation (37) is the SOP capacity constraint, and Equation (38) indicates that if one end of the SOP is located on the power outage side, it is adjusted to the V f control mode;

[0183] In the step S3, the double-layer optimization and reconstruction model includes the following parts:

[0184] S3-1: The upper-layer objective function is in the following form:

[0185]

[0186] Among them, α is the power outage cost coefficient, β is the cost coefficient of wind and light abandonment, and γ is the SOP loss cost coefficient;

[0187] The goal of the upper-layer model is to complete the planning of the switching operations of the distribution network switches, OLTC operations, and capacitor bank switching every hour at the moment and in the subsequent time after a fault occurs, taking into account the randomness of the load and photovoltaic power generation. S3-2: The objective function form of the lower-layer model is as follows:

[0188]

[0189] The goal of the lower-layer model is to obtain the optimal solutions for the DG output, load shedding size, and network and SOP losses within the next 15 minutes based on the current load and photovoltaic power forecasts, on the basis of determining the actions of the above-mentioned devices, so as to maximize the restoration of the power supply capacity of the power grid and reduce system losses.

[0190] To enable those skilled in the art to better understand the present invention, the case study includes the following components:

[0191] I. Case study description and simulation result analysis

[0192] In order to verify its effectiveness, the present invention uses the system shown in Figure 2 for case study analysis. Three photovoltaic generators are placed at nodes 6, 10, and 27 respectively, and two wind turbines are placed at 13 and 30. The total output of the new energy devices is as shown in Figure 3 . The reference voltage is set to 12.66 KV, the voltage amplitude boundary is [0.95, 1.05] p.u., the maximum transmission power on the line is set to 4 MW. The SOP capacity upper limit is 0.8 MVA, the SOP loss coefficient is 0.01, the power outage cost coefficient in the upper and lower layer objective functions is 40, the curtailment cost coefficient of wind and light is 1, and the SOP loss cost coefficient is 2. CVR p and CVR q are 0.7692 and 2.2154 respectively.

[0193] To fully demonstrate the effectiveness of the proposed method, another mode is set for comparative simulation analysis:

[0194] 1) Mode 1: The fault recovery strategy with tie switches and SOP proposed in the present invention.

[0195] 2) Mode 2: The fault recovery strategy without SOP and only with tie switches.

[0196] The simulation program is implemented in the Matlab environment of a computer with Windows10, Intel(R) CoreTM i5 CPU@3.5GHz, and 8GB memory. First, the action voltage values of the OLTC in Table 1 are obtained. Taking the 17th moment as an example, the system voltage distribution as shown in Figure 4 is obtained. From Figure 4It can be seen that the two peak points are at node 18 and node 27. Node 18 changes from SOP to V f Control is performed for regional voltage support. Node 27 is the main generator at this location, and the SOP at node 29 raises the voltage to the rated voltage. Figure 5 Indicates the losses of the system at each time period, Figure 6 Indicates the voltage distribution of the system and its change over time, Figure 7 Indicates the error magnitude of the second-order cone conversion at different nodes at different times. Calculate and compare the power outage rate, light energy absorption rate, wind energy absorption rate, maximum voltage, minimum voltage, and voltage deviation in the system under the above two operating modes. The specific statistical data is shown in Table 2.

[0197] Table 1 OLTC operating voltage values

[0198]

[0199] Table 2 Comparison diagram of the two scenarios of the system

[0200]

[0201] As can be seen from Table 2, due to the precise power flow regulation ability of SOP, Mode 1 can reduce the power outage rate of the system, improve the light energy and wind energy absorption rates, and reduce the voltage deviation. Under the action of SOP, the power outage rate of Mode 1 is reduced by 9.02%, the light energy absorption rate is increased by 43.86%, and the wind energy absorption rate is increased by 33.6%. Thus, it can be seen that the SOP fault recovery strategy proposed in this invention can improve the recovery performance of the distribution system in the fault scenario.

[0202] In this specification, the schematic expressions of the present invention are not necessarily directed to the same embodiments or examples. Those skilled in the art can combine and combine the different embodiments or examples described in this specification. In addition, the content described in the embodiments of this specification is only a list of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms stated in the implementation cases. The protection scope of the present invention also includes equivalent technical means that those skilled in the art can think of according to the inventive concept of the present invention.

Claims

1. A rolling optimization method for distribution network fault recovery considering the randomness of load photovoltaics, characterized in that: The following steps are involved: S1: Construct a photovoltaic load randomness model, using Gaussian distribution to describe the random prediction error of load and normal distribution to describe the random prediction error of photovoltaic power output; S2: Construct a power flow constraint model, a node ZIP load constraint model, an on-load tap changer (OLTC) control model, a topology and virtual power flow constraint model, and an intelligent soft open point (SOP) model. S3: Construct a two-layer optimization and reconstruction model. The upper layer of the model uses the distribution network architecture, the operation of the on-load tap-changing transformer, and the switching of the capacitor bank as decision variables. The lower layer optimization objectives are the active and reactive output of distributed energy, the power of load shedding, and the size of the system network loss. By setting the same mathematical form of objective functions, the objective functions are optimized for the entire period and the subsequent period respectively, to achieve distribution network fault reconstruction and power flow optimization. The two-layer optimization and reconstruction model includes the following parts: S3-1: The upper objective function is as follows: Among them, α is the power loss cost coefficient, β is the wind and solar curtailment cost coefficient, and γ is the SOP loss cost coefficient; The upper-level model aims to plan the distribution network switch actions, OLTC actions, and capacitor bank switching every hour at the moment of fault occurrence and shortly thereafter, taking into account the randomness of load and PV generation. S3-2: The objective function of the lower model is as follows: The goal of the lower-level model is to determine the optimal solution for DG output, load shedding, and network and SOP losses within the next 15 minutes based on the current load and photovoltaic forecast values, based on the determined device actions. This will maximize the grid's power supply capacity and minimize system losses. S4: Obtain the relevant parameters required for optimization and solve the model.

2. A distribution network fault recovery rolling optimization method considering load photovoltaic randomness according to claim 1, characterized in that: Step S1 specifically includes: S1-1: The daily load data of the distribution network is obtained from the load forecast. The Gaussian distribution is used to describe the random forecast error of the load. The probability density function of the load data is: Among them, P LD , Q LD is the load active and reactive output, μ LD is the output prediction value, σ LD is the standard deviation of the prediction error; 500 sets of random output scenarios are generated using this probability density function, and then reduced to 20 sets of scenarios through backward reduction for use in subsequent examples; S1-2: Photovoltaic output can be predicted based on meteorological conditions. The normal distribution is used to describe the random prediction error of photovoltaic power output. The probability density function of photovoltaic active output is: Among them, P PV is the photovoltaic active power output, μ PV is the output prediction value, σ PV is the standard deviation of the prediction error. 500 sets of random output scenarios are generated using this probability density function, and are reduced to 20 sets of scenarios through backward reduction for use in subsequent examples.

3. A distribution network fault recovery rolling optimization method considering load photovoltaic randomness according to claim 1, characterized in that: The construction of the power flow constraint model, the node ZIP load constraint model, the OLTC control model, the topology and virtual power flow constraint model, and the SOP model described in step S2 specifically includes: S2-1: The distribution network power flow model is a power flow equation established based on branch power. Compared with the traditional power flow calculation based on node power, the power flow model is more suitable for power flow calculation of radial distribution systems. In the distribution network reconstruction problem, due to the problem of changes in branch disconnection conditions, it is necessary to improve the traditional power flow model and introduce the line disconnection variable Z. ij By relaxing the power flow equation and selecting the π-type equivalent line model, the following power flow equation suitable for distribution network reconstruction is obtained: Among them, Ω s For all conducting branches, r ij and x ij is the resistance and reactance of branch ij, is the admittance of branch ij, Z ij is the breaking variable of branch ij; P ij,ω,t is the active power flowing from node i to node j on the branch; Q ij,ω,t is the reactive power flowing from node i to node j on the branch; is the square value of the current flowing from node i to node j on the branch; is the square value of the voltage at node i; and is the node’s power-off active and reactive power; is the active output of the SOP port at node i, is the reactive power output of the SOP port at node i; and are the active power injected by the distributed generation and consumed by the load at node i respectively; is the photovoltaic power generation power at the node, is the photovoltaic curtailment power at the node; Q i,ω,t is the sum of reactive power injected into node i, P i,ω,t is the sum of active power injected into node i; and are the reactive power injected by the distributed generation and consumed by the load at node i respectively; is the reactive power provided by the capacitor bank; The upper and lower limits of active and reactive output of various types of equipment are as follows: Line power constraints: Node voltage and current constraints: Power loss power constraint: Generator constraints: Wind and solar curtailment constraints: in, P ji , Represent the minimum and maximum active power of the line respectively, Q ji , Respectively represent the minimum and maximum reactive power of the line; U i , Represent the minimum and maximum node voltages respectively; Respectively represent the maximum active and reactive power of the generator; S2-2: The most widely accepted distribution network load model at this stage is the ZIP model, which divides the load power demand into three parts: constant impedance Z, constant current I, and constant power P. The load can be expressed as follows: Among them, k p,1 +k p,2 +k p,3 =1,k q,1 +k q,2 +k q,3 =1, Rated voltage U N Rated active power value under ; Rated voltage U N Rated reactive power value under ; Node power is a nonlinear function of node voltage and cannot be incorporated into the linear programming model used in this paper. CVR is defined as the ratio of the percentage of active or reactive power to the percentage of bus voltage reduction. The definition of CVR yields an equivalent model for voltage-sensitive loads, as follows: in, There are Considering U i,ω,t ≈U N and get: At this time, formulas (18) and (19) have been linearized and included in the power flow constraints, and the value of CVR is estimated from the ZIP coefficient of the load; the original model is examined and U is set N =1p.u., we get: By formula (24), (25) and assuming U i,ω,t ≈U N , the CVR value is as follows: CVR P =2k P，1 +k P，2 (26) CVR Q =2k Q，1 +k Q，2 (27) S2-3: OLTC adjusts the output voltage within a certain range by adjusting the tap position. The model is established as follows: in, represents the position of the tap at time t, is the square value of the voltage corresponding to the tap position, is the square of the OLTC output voltage, represents the square of the per-unit value of the voltage at node 1 at time t, U Respectively represent the upper and lower limits of the node voltage; S2-4: Distribution network reconstruction must ensure the connectivity of the reconstructed distribution system, without islands or rings. Therefore, the connectivity and radiation constraints of the distribution system can be expressed as: X 12，t =z 12，t (32) Where E is the set of line nodes, z ij,t is the circuit breaking variable, X ij,t For parent and child nodes, X represents variables. ij,t =1 means that at time t, i is the parent node of j, λ j,t is the virtual power demand of the node, is the virtual power supply of the generator, is the line virtual power, c(j), δ(j) represents the adjacent nodes of j, Ω DG represents the set of nodes connected to the distributed generator, Ω SOP Represents the set of nodes connected to SOP, Ω G Represents the set of nodes connected to the upper power grid; Formulas (33)-(35) indicate that a node has at most one parent node, and its parent node must be greater than its own virtual power flow demand; S2-5: The SOP is a new type of intelligent distribution device that replaces traditional tie switches. Its application will greatly improve the flexibility and controllability of distribution system operation. However, the role of the SOP in the self-healing process of distribution network faults has been less studied. Compared with tie switches, the power control of the SOP is more accurate and reliable, avoiding the potential safety hazards caused by switch operation. When a fault occurs, the DC isolation function can effectively prevent the fault current from passing through. During the power supply restoration process, it can provide effective voltage support for the power-off side, thereby expanding the power supply restoration range. Currently, the back-to-back voltage source converter (B2B VSC) type SOP is a relatively common SOP. Its topology is realized by two converters connected through a DC capacitor. Adding SOPs to distribution system branches improves power flow distribution, reduces system losses, and balances network voltage. After a distribution network fault occurs, a power outage area is formed after fault location and isolation. However, as a distribution device that replaces the tie switch, the SOP cannot restore power to any power outage area in the entire distribution system. The following article analyzes power outage areas that can be restored using the SOP. After a distribution network fault occurs, the power-off area formed by fault location and fault isolation can be divided into the following three situations based on the SOP access location: 1) Both ends of the SOP are not in the power-off area; 2) One end of the SOP is in the power-off area, while the other end is not in the power-off area; 3) Both ends of the SOP are in the power-off area; When a certain area can be connected to the upper network through the tie switch, SOP runs in PQ control mode and optimizes its active and reactive outputs; when the power-off area cannot be connected to the upper network through the tie switch, SOP adopts V f In the control mode, the outlet voltage of the distribution network SOP and the switch status in the distribution network should be adjusted to recover from the fault, and the remaining unrecovered loads form corresponding islands. Considering the two scenarios 1 and 2, the SOP constraints are proposed as follows: Voltage constraint on the SOP power-off side: in, is the loss coefficient of SOP; is the SOP loss connected to node i; is the SOP capacity connected to node i; Ω n represents the set of nodes on the power-off side; U0 is the minimum per-unit voltage limit of the nodes on the power-off side, which is generally 1.0; formula (36) is the SOP active power constraint, formula (37) is the SOP loss constraint, formula (38) is the SOP capacity constraint, and formula (39) indicates that if one end of the SOP is on the power-off side, it is adjusted to V f Control mode.

4. A distribution network fault recovery rolling optimization method considering load photovoltaic randomness according to claim 1, characterized in that: The objective functions described in step S3 are full-time optimization and subsequent-time optimization.

Images