A power distribution network recovery method based on photovoltaic power generation controllability recovery power

Abstract

The application discloses a power distribution network recovery method based on photovoltaic power generation controllability recovery power, and the method comprises the following steps: acquiring meteorological information and photovoltaic equipment information of a prediction day, and predicting photovoltaic output by adopting a photovoltaic output prediction method based on multi-factor correction; generating a photovoltaic output prediction curve according to the photovoltaic output prediction result, decomposing the photovoltaic output prediction curve, and decomposing a photovoltaic power station into a plurality of virtual power source combinations which have specific time period availability and controllable output power; selectively recovering un-recovered electric charges by adopting constraint conditions according to the power distribution network and virtual power source data, constructing a target function and solving the target function to obtain island division results and load recovery sequences; and recovering the power distribution network according to the island division results and the load recovery sequences. The application reduces the influence of photovoltaic output fluctuation on power system recovery, and makes the system more stable and reliable.

Description

Technical Field

[0001] This invention belongs to the field of power system restoration, specifically, it relates to a method for restoring distribution network output based on the controllability of photovoltaic power generation. Background Technology

[0002] The distribution network is a crucial link in the power grid, directly connected to and distributing electrical energy to consumers. Its stable operation is essential for maintaining normal work and life for residents, social stability, and ensuring energy security. As renewable energy generation accounts for an increasingly larger proportion of the power system's output, wind and solar power generation cannot provide the long-term, stable energy required for recovery like fossil fuel power generation. Their strong volatility and uncertainty not only increase the difficulty of distribution network recovery but, more seriously, can lead to further collapses during system recovery. Therefore, research considering a high proportion of renewable energy participation in power system recovery can adapt to the increasing proportion of renewable energy generation in the system and effectively improve the resilience of the power system.

[0003] With the large-scale integration of new energy sources into the power grid, research considering the participation of photovoltaic (PV) power generation in distribution network restoration has also developed. Patent publication CN103078391A proposes a distribution network power restoration method based on a PV power generation system. This method mainly focuses on the location of the PV system and addresses the islanded operation of the PV system location, but it does not fully consider the actual participation of PV power output in distribution network restoration. This leads to a waste of PV power output in practical applications, failing to fully utilize the potential role of the PV system in the power system restoration process.

[0004] Therefore, a distribution network restoration method that comprehensively considers the actual participation of photovoltaic power output needs to be proposed. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a distribution network restoration method based on the controllability of photovoltaic power generation.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] A method for restoring distribution network output based on the controllability of photovoltaic power generation, characterized in that it includes:

[0008] Step 1: Obtain meteorological information and photovoltaic equipment information for the forecast date, and use a photovoltaic output forecasting method based on multi-factor correction to forecast photovoltaic output;

[0009] Step 2: Generate a photovoltaic output prediction curve based on the photovoltaic output prediction results, decompose the photovoltaic output prediction curve, and decompose the photovoltaic power station into multiple virtual distributed power combinations that are available for specific time periods and have controllable output power.

[0010] Step 3: Based on the distribution network data, selectively restore unrestored charges using constraints, construct an objective function and solve it to obtain the unrestored loads and load restoration order in the islanding partitioning results. The distribution network data includes the virtual distributed power source data.

[0011] Step 4: Restore the distribution network according to the islanding results and load restoration order.

[0012] Further, step 1 includes:

[0013] Establish a physical model for photovoltaic output, obtain meteorological and equipment information for the forecast date, and determine the preliminary photovoltaic power forecast for the forecast date;

[0014] The factors influencing photovoltaic power output prediction were identified, and the importance weight vector of the influencing factors corresponding to each variable importance assessment method was obtained by using multiple variable importance assessment methods.

[0015] The fuzzy multi-criteria method is used to evaluate the various variable importance assessment methods, and the optimal assessment method is determined from among the various variable importance assessment methods.

[0016] The correction coefficient for photovoltaic output is determined based on the weight vector of the importance of the influencing factors corresponding to the optimal evaluation method.

[0017] The final photovoltaic power forecast is determined based on the correction factor and the preliminary photovoltaic power forecast.

[0018] Furthermore, the multiple variable importance assessment methods include at least two of the following: a variable importance assessment method based on the RF algorithm and Gini coefficient, a variable importance assessment method based on the reduction of average precision, and a variable importance assessment method based on the entropy weight method.

[0019] Furthermore, a fuzzy multi-criteria method is used to evaluate the importance assessment methods for the multiple variables, including:

[0020] Using the determined influencing factors of photovoltaic power output prediction and the weights of the influencing factors corresponding to the importance assessment method of each variable, a weight matrix H is constructed;

[0021] The optimal evaluation method is determined based on the weight matrix H, expressed by the formula:

[0022] R = λY = [r1 r2 … r a ];

[0023] in,

[0024] λ=[λ1 λ2 … λ n ],

[0025]

[0026]

[0027]

[0028]

[0029] Where n represents the number of influencing factors; R represents the decision vector; and r1, r2, and r3 represent the decision vectors. a Let r1, r2, and r3 represent the decision values ​​corresponding to each method for assessing the importance of variables, respectively. The optimal assessment method, r1, r2, and r3, is determined at the end. a The maximum value in the value corresponds to the variable importance assessment method, where 'a' represents the number of variable importance assessment methods participating in the fuzzy multi-criteria method; λ represents the weight vector; δ represents the parameter representing the number of decision variables; h ij This represents the importance value of the i-th influencing factor obtained by the j-th variable importance assessment method.

[0030] Furthermore, based on the importance weight vector of the influencing factors corresponding to the optimal evaluation method, the correction coefficient for photovoltaic output is determined, including:

[0031] Assuming the optimal evaluation method corresponds to the influence factor importance weight vector λ', and the predicted value x for each influence factor is obtained, then;

[0032] Based on the predicted value x and weight vector λ' for each influencing factor, the comprehensive prediction score S is obtained;

[0033] The correction coefficient is determined based on the comprehensive prediction score.

[0034] Furthermore, based on the predicted value x and weight vector λ' for each influencing factor, the comprehensive prediction score S is obtained, expressed by the formula:

[0035]

[0036] Where x = [x1, x2, x3, ..., x n ], x i λ' represents the predicted value corresponding to the i-th influencing factor; λ' = [λ'1, λ'2, λ'3, ..., λ'] n ], λ i ' represents the importance weight of the i-th influencing factor obtained by using the optimal evaluation method, and n represents the number of influencing factors.

[0037] Furthermore, the correction coefficient is determined based on the comprehensive prediction score, expressed by the formula:

[0038]

[0039] in:

[0040]

[0041]

[0042] k0 and b represent the parameters set, x max and x min These represent the upper and lower limits of the value range for the influencing factors.

[0043] Furthermore, the decomposition of the photovoltaic output prediction curve in step 2 includes:

[0044] Step 31: Extract the node data of the photovoltaic power output prediction curve and calculate the variance σ of the node data, wherein the node data represents the photovoltaic output power prediction results obtained at each time point;

[0045] Step 32: Set the constant sequence U = [U1, U2, ..., U e ], where let s = (σ * 10) 2 After rounding s to an even number, we get e. We then use the sequence U to partition the horizontal axis of the photovoltaic power output prediction curve.

[0046] Step 33: Set the point with the minimum slope between two points in each interval. This is the decomposition point;

[0047] Step 34: and and and Decomposition points are paired up to create decomposition pairs;

[0048] Step 35: Perform rectangular decomposition on the decomposed pairs to obtain the photovoltaic output virtual distributed power source.

[0049] Furthermore, the objective function in step 3 is expressed by the formula:

[0050]

[0051]

[0052] In the formula: T represents the total set of the recovery phase; L represents the set of distribution network nodes; Δt represents the duration of the recovery phase; P i',k' γ represents the active power of the load on node i' within island k' formed after the fault; i',k' Represents the node weight coefficient; it is a binary search. Variables, where The load representing node i' is waiting for power to be restored. The load at node i' has been restored to power; S represents the collection of photovoltaic power generation devices and energy storage devices; This represents the network loss of the entire system during stage t. This represents the power output of the photovoltaic power generation device j' during stage t; denoted by t+1, where is the power output of photovoltaic power generation device j' in stage t+1; w is the weighting coefficient of the objective function; and K represents the island set.

[0053] Furthermore, the constraint function in step 3 includes:

[0054] (1) Capacity constraint:

[0055]

[0056] In the formula, This represents the remaining capacity of photovoltaic power generation devices and energy storage devices within island k';

[0057] (2) Energy storage constraints:

[0058]

[0059] In the formula, and Let J' represent the remaining power generation capacity and discharge power of the energy storage device j' on the node within island k' at time t, respectively. and Let represent the maximum capacity, minimum capacity, maximum discharge power, and minimum discharge power of the energy storage devices within island k', respectively; G represents the collection of energy storage devices.

[0060] (3) Node voltage constraints:

[0061] U tmin ≤U i' ≤U tmax

[0062] In the formula: U tmin U represents the voltage amplitude at node i. i' The lower limit of U; tmax This indicates the upper limit of the voltage amplitude at node i;

[0063] (4) Network radial constraints:

[0064] c∈C

[0065] In the formula: c represents the area where power is restored; C represents the set of radial topologies of the guaranteed network;

[0066] (5) Node power balance constraints:

[0067]

[0068] In the formula: within the isolated island k' and These represent the active power and reactive power flowing from node i' to node g' on line (i', g'), respectively. and This represents the active and reactive power of the load connected to node g'; and The active and reactive power injected into the distributed power sources on node g'; r i'g' and x i'g' These are the resistance and reactance on the line (i', g'), respectively; Let be the square of the current amplitude on line (i', g').

[0069] The beneficial effects of this invention are:

[0070] This invention first employs a dual-driven approach based on physical models and data to accurately predict photovoltaic (PV) power output. Then, by using a multi-period modeling method for the controllable recovery characteristics of PV power output, the PV power generation system with long-term output fluctuations is transformed into multiple power sources with locally controllable output, thus participating in system recovery decisions. This model can be applied online to optimize the power distribution network recovery process for systems with a high proportion of PV power generation. This invention optimizes the fluctuating PV power output through morphological decomposition to obtain controllable PV power output. This optimization reduces the impact of PV power output fluctuations on power system recovery, making the system more stable and reliable.

[0071] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0072] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will now be described in further detail with reference to the accompanying drawings, wherein:

[0073] Figure 1 This is a schematic flowchart of a distribution network restoration method based on the controllability of photovoltaic power generation.

[0074] Figure 2 This is an exemplary flowchart of a photovoltaic power output prediction method based on multi-factor correction;

[0075] Figure 3 This is a schematic diagram of the output curve decomposition of a photovoltaic power station;

[0076] Figure 4 This is a diagram of the IEEE 33-node system;

[0077] Figure 5 This is a simulation result diagram of the node voltage;

[0078] Figure 6 This is a graph showing the simulation results of the load recovery rate. Detailed Implementation

[0079] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the preferred embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0080] Based on the dual-drive photovoltaic output prediction method, this application proposes a photovoltaic power generation controllability recovery output model that considers the multi-period decomposition of the photovoltaic curve. Furthermore, combined with the distribution network recovery model, a distribution network recovery strategy that considers the multi-period output of photovoltaic power is proposed.

[0081] Figure 1 A schematic flowchart illustrating a distribution network restoration method based on the controllability of photovoltaic power generation. (See attached diagram.) Figure 1 As shown, the method includes:

[0082] Step 1: Obtain meteorological and photovoltaic (PV) equipment information for the forecast date, and use a PV output forecasting method based on multi-factor correction to predict PV output. For details of this step, please refer to [link to relevant documentation]. Figure 2 And its related descriptions.

[0083] Step 2: Generate a photovoltaic output prediction curve based on the photovoltaic output prediction results, decompose the photovoltaic output prediction curve, and decompose the photovoltaic power station into multiple virtual distributed power combinations that are available for specific time periods and have controllable output power.

[0084] Step 3: Based on the distribution network data, selectively restore unrestored loads using constraints, construct an objective function and solve it to obtain the islanding results and load restoration order; wherein, the distribution network data includes the virtual distributed power source data;

[0085] Step 4: Restore the distribution network according to the islanding results and load restoration order.

[0086] Figure 2 This is an exemplary flowchart of a photovoltaic power output prediction method based on multi-factor correction. Figure 2 As shown, the method includes the following steps:

[0087] Step 11: Establish a physical model for photovoltaic output, obtain meteorological and equipment information for the forecast date, and determine the preliminary photovoltaic power forecast for the forecast date;

[0088] Step 22: Determine the influencing factors for photovoltaic power output prediction, and obtain the influence factor importance weight vector corresponding to each variable importance assessment method using multiple variable importance assessment methods;

[0089] Step 33: Evaluate the multiple variable importance assessment methods using the fuzzy multi-criteria method, and determine the optimal assessment method from among the multiple variable importance assessment methods;

[0090] Step 44: Determine the correction coefficient for photovoltaic output based on the importance weight vector of the influencing factors corresponding to the optimal evaluation method;

[0091] Step 55: Determine the final photovoltaic power forecast based on the correction coefficient and the preliminary photovoltaic power forecast.

[0092] In step 11, the physical model for photovoltaic power output is as follows:

[0093] P pyh =(a1+a2I+a3ln I)(1-β(T) C -T0))AI,

[0094] Among them, P pyh Indicates the total predicted photovoltaic power; I represents the solar irradiance; A represents the photovoltaic array area; T C The temperature is represented by a1, a2, and a3, which are photovoltaic equipment parameters obtained by fitting historical data using the least squares method. β and T0 represent the set parameters. As an example, β can be 0.005, and T0 can be 25 degrees Celsius.

[0095] The physical model described above considers the main factors affecting photovoltaic (PV) output, such as solar irradiance, current temperature, and PV array area, and models them mathematically. After the model is established, historical data (e.g., including historical solar radiation, temperature, cloud cover, rainfall, and corresponding PV output power) is obtained and fitted (e.g., using the least squares method) to obtain PV equipment parameters. These parameters are then used in conjunction with forecast data on solar radiation, temperature, cloud cover, and rainfall for the predicted day, along with other PV equipment parameters (e.g., PV array area), to predict the PV output power for the predicted day, thus obtaining a preliminary PV power prediction.

[0096] In step 22, the influencing factors for photovoltaic (PV) power output prediction are determined. These factors may include equipment information, meteorological information, and altitude / geographical information. Equipment information may include PV capacity, PV panel angle, and conversion efficiency. Meteorological information may include solar radiation, cloud cover, rainfall, and temperature. These features can be extracted from historical samples using variable feature weighting methods. Then, various variable importance assessment methods can be used to obtain the influence factor importance weight vector corresponding to each method.

[0097] In some embodiments, the multiple variable importance assessment methods include at least two of the following: a variable importance assessment method based on the RF algorithm and Gini coefficient, a variable importance assessment method based on the reduction of average precision, and a variable importance assessment method based on the entropy weight method.

[0098] The variable importance assessment method based on the RF algorithm and Gini coefficient can be described as follows:

[0099] Suppose we have the following features: photovoltaic capacity, photovoltaic panel angle, conversion efficiency, solar radiation, and rainfall, corresponding to the variables X1, X2, X3, ..., X... n Where n represents the number of influencing factors. Based on the RF (Random Forest) algorithm, the relationship between data-driven factors and the output power of the physical model is treated as a dichotomous variable, i.e., greater than or equal to or less than. The variable X is calculated based on the Gini coefficient. i The weights.

[0100] The Gini coefficient can be expressed as:

[0101] G m =2P m (1-P m ),

[0102] In the formula, P m This is an estimate of the probability that a sample belongs to any class at node m.

[0103] VARIABLE X i Importance of node m V im That is, the change in the Gini coefficient before and after the branch at node m, i.e.:

[0104] V im =G m -G l -G s ,

[0105] In the formula: G l and G s are the Gini coefficients of the two new nodes l and s that split from node m.

[0106] If variable X i If X appears q times in the j-th tree, then the variable X... i The importance of the j-th tree is:

[0107]

[0108] VARIABLE X i In RF, Gini importance is defined as:

[0109]

[0110] In the formula: r represents the number of classification trees in RF, that is, the number of influencing factors.

[0111] By analyzing each variable (i.e., influencing factor) X i Calculate the importance weight V i These weights can form a weight vector.

[0112] Mean precision is a commonly used feature selection method that directly evaluates the impact of each feature on the model's precision. The main idea is to shuffle the order of feature values ​​and assess the impact of this order change on model precision. For unimportant variables, shuffling the order has little impact on model precision; however, for important variables, shuffling the order will reduce model precision. A variable importance assessment method based on reduced mean precision can be described as follows:

[0113] The sample data is divided into a training set and a test set, with 80% being training samples and 20% being test samples.

[0114] Using the RF model, the RF model is trained and the parameters are adjusted to obtain the accuracy g;

[0115] Calculate variable X i Impact on prediction accuracy. Let variable X... i The corresponding feature data is randomly shuffled c times. Calculate the test accuracy S after the j-th shuffle. ji (i = 1, 2, 3, ..., n), then variable X i The effect of the j-th shuffling on test accuracy S ji As shown in the following formula:

[0116]

[0117] In the above formula, g ji ′(j=1,2,3…,n) represents the test accuracy after the i-th shuffle, which can be determined by the variable X. i The corresponding feature data is obtained by randomly shuffling it n times.

[0118] VARIABLE X iImpact on test accuracy S i for:

[0119]

[0120] Next, the importance weights ω for each variable are calculated. i ,Right now

[0121]

[0122] By analyzing each variable (i.e., influencing factor) X i Calculate the importance weight ω i These weights ω i This can form a weight vector.

[0123] Entropy weighting is based on objective data analysis, resulting in relatively objective conclusions and a simpler calculation process, making it widely used in multi-factor weighting analysis. In entropy weighting, the information entropy of each indicator is negatively correlated with the degree of difference in its values; the greater the difference in values, the smaller the information entropy, the greater the amount of information, and the larger the final weight.

[0124] The variable importance assessment method based on entropy weight can be described by the following formula:

[0125]

[0126] In the formula: n represents the number of variables; f represents the number of samples; D is the evaluation matrix; D ij P represents the i-th similarity day and the j-th variable in the evaluation matrix; ij For variable X j The weight of the j-th sample in the sample; e i For variable X i entropy; b i The entropy weight is used.

[0127] The variable importance assessment method based on entropy weighting yields each variable X... i The corresponding entropy weight b i This is the importance weight of the variable. The importance weights of n variables can form a weight vector.

[0128] In step 33, the fuzzy multi-criteria method is used to evaluate the importance assessment method of the multiple variables, including:

[0129] Using the determined influencing factors of photovoltaic power output prediction and the weights of the influencing factors corresponding to the importance assessment method of each variable, a weight matrix H is constructed;

[0130] The optimal evaluation method is determined based on the weight matrix H, expressed by the formula:

[0131] R = λY = [r1 r2 … r a ];

[0132] in,

[0133] λ=[λ1 λ2 … λ n ],

[0134]

[0135]

[0136]

[0137]

[0138] Where n represents the number of influencing factors; R represents the decision vector; and r1, r2, and r3 represent the decision vectors. a Let r1, r2, and r3 represent the decision values ​​corresponding to each method for assessing the importance of variables, respectively. The optimal assessment method, r1, r2, and r3, is determined at the end. a The maximum value in the fuzzy multi-criteria method corresponds to the variable importance assessment method, where 'a' represents the number of variable importance assessment methods involved in the fuzzy multi-criteria method; λ represents the weight vector; δ represents the parameter for the number of decision variables, and since the elements in the weight vector are all undetermined, δ is set to 1; h ij This represents the importance value of the j-th variable to the i-th influencing factor, obtained by the j-th variable importance assessment method.

[0139] In step 44, based on the importance weight vector of influencing factors corresponding to the optimal evaluation method, the correction coefficient for photovoltaic output is determined, including:

[0140] Assume that the optimal evaluation method corresponds to the weight vector λ' of the importance of the influencing factors, and the predicted value x for each influencing factor;

[0141] Based on the predicted value x and weight vector λ' for each influencing factor, the comprehensive prediction score S is obtained;

[0142] The correction coefficient is determined based on the comprehensive prediction score.

[0143] Among them, the comprehensive prediction score S is obtained based on the predicted value x and weight vector λ' corresponding to each influencing factor, which is expressed by the formula:

[0144]

[0145] Where x = [x1, x2, x3, ..., x n ], x i λ' represents the predicted value corresponding to the i-th influencing factor; λ' = [λ'1, λ'2, λ'3, ..., λ'] n ], λi ' represents the importance weight of the i-th influencing factor obtained by using the optimal evaluation method.

[0146] Furthermore, the correction coefficient is determined based on the comprehensive prediction score, expressed by the formula:

[0147]

[0148] Considering that the increase in the values ​​of various influencing factors has different effects on the scoring driver, such as the positive correlation between light intensity, array area and photovoltaic output power, while the negative correlation between ambient temperature, operating years, and cloud cover and photovoltaic output power, therefore:

[0149]

[0150]

[0151] k0 and b represent the parameters set; for example only, k0 can be 0.4 and b can be 0.9. max and x min These represent the upper and lower limits of the value range for the influencing factors.

[0152] In step 55, based on the correction coefficient and the preliminary photovoltaic power forecast, the final photovoltaic power forecast is determined using the following formula:

[0153] P = kP pyh ,

[0154] Where P represents the final photovoltaic power prediction.

[0155] pass Figure 2 The method shown can predict the photovoltaic output power for any time period of the forecast day. When predicting any time period, the preliminary photovoltaic power forecast for any time period can be calculated first. Then, the correction coefficient for that arbitrary time period can be calculated using the method described in steps 22-44 of this application. Finally, the predicted photovoltaic output power value for any time period can be obtained according to step 55.

[0156] By predicting photovoltaic (PV) output, the mechanism by which PV power generation participates in power restoration can be analyzed based on the characteristics of power system recovery. Unlike power system flexibility, which requires maximizing the absorption of new energy sources, power system resilience prioritizes system safety. Because the power system is relatively vulnerable during recovery and may experience secondary power outages due to strong fluctuations, the power range with high volatility (i.e., the uncontrollable portion) of PV output should be discarded, while ensuring that the controllable output range remains to participate in power restoration (i.e., the controllable portion).

[0157] Based on the predicted photovoltaic output power for any time period on the forecast date, a photovoltaic power output prediction curve can be obtained. By decomposing the curve shape, the photovoltaic power station can be decomposed into multiple virtual power source combinations with specific time periods of availability and controllable output power. Figure 3 This is a schematic diagram of the output curve decomposition of a photovoltaic power station. (For example...) Figure 3 As shown, rectangle S (including S1-S4) represents the controllable output power of the virtual power source after the output curve decomposition within a specific time period. Using the power range as the decomposition standard, the decomposed virtual power sources have a coupling relationship in the time interval. However, the number of variables in the decomposed virtual power sources is relatively small, and the controllable recoverable output range of each virtual power source is similar, making it suitable for curves with relatively gentle fluctuations.

[0158] To address the specific characteristics of photovoltaic (PV) power plant output prediction curves, a lateral decomposition method based on curve shape is selected to decompose the PV power plant output prediction curves. In this invention, based on the volatility of the curve, the more volatile the curve, the more decomposition points are set. Each PV unit is designated as a virtual power source, and the power of each virtual power source is obtained by adding the sub-powers within each decomposed interval. The number of virtual power sources is determined by the partitioning; the number of intervals (i.e., the "decomposition pairs" mentioned below) determines the number of virtual power sources. The specific decomposition steps are as follows:

[0159] Step 31: Extract the node data of the photovoltaic power output prediction curve and calculate the variance σ of the node data, wherein the node data represents the photovoltaic output power prediction results obtained at each time point;

[0160] Step 32: Set the constant sequence U = [U1, U2, ..., U e ], where let s = (σ * 10) 2 After rounding s to an even number, we get e. We then use the sequence U to partition the horizontal axis of the photovoltaic power output prediction curve.

[0161] Step 33: Set the point with the minimum slope between two points in each interval. This is the decomposition point;

[0162] Step 34: and and and Decomposition points are paired up to create decomposition pairs;

[0163] Step 35: Perform rectangular decomposition on the decomposed pairs to obtain the photovoltaic output virtual distributed power source.

[0164] refer to Figure 3 Taking virtual power sources (i.e., virtual distributed power sources) S3 and S4 as examples, the horizontal decomposition of the output curve shape can be expressed as follows:

[0165]

[0166] Among them, P S3 and P S4 P1 and P4 represent the power of virtual power sources S3 and S4, respectively; P3 and P4 represent the upper limits of the power of virtual power sources S3 and S4, respectively; t1 to t8 represent exemplary decomposition points.

[0167] Returning to step 4, this invention combines a multi-period controllable output characteristic model of photovoltaic power generation to ensure as much conformity recovery as possible during the recovery process. This prevents secondary power outages caused by strong fluctuations in photovoltaic output during distribution network recovery, thereby reducing distribution network losses. The objective function for distribution network recovery proposed in this invention is:

[0168]

[0169]

[0170] In the formula: T represents the total set of the recovery phase; L represents the set of distribution network nodes; Δt represents the duration of the recovery phase; P i',k' γ represents the active power of the load on node i' within island k' formed after the fault; i',k' Represents the node weight coefficient; it is a binary search. Variables, where The load representing node i' is waiting for power to be restored. The load at node i' has been restored to power; S represents the collection of photovoltaic power generation devices and energy storage devices; This represents the network loss of the entire system during stage t. This represents the power output of the photovoltaic power generation device j' during stage t; denoted by t+1, where is the power output of photovoltaic power generation device j' in stage t+1; w is the weighting coefficient of the objective function; and K represents the island set.

[0171] The constraint functions include:

[0172] (1) Capacity constraint:

[0173]

[0174] In the formula, This represents the remaining capacity of photovoltaic power generation devices and energy storage devices within island k';

[0175] (2) Energy storage constraints:

[0176]

[0177] In the formula, and Let J' represent the remaining power generation capacity and discharge power of the energy storage device j' on the node within island k' at time t, respectively. and Let represent the maximum capacity, minimum capacity, maximum discharge power, and minimum discharge power of the energy storage devices within island k', respectively; G represents the collection of energy storage devices.

[0178] (3) Node voltage constraints:

[0179] U tmin ≤U i' ≤U tmax

[0180] In the formula: U tmin U represents the voltage amplitude at node i. i' The lower limit of U; tmax This indicates the upper limit of the voltage amplitude at node i;

[0181] (4) Network radial constraints:

[0182] c∈C

[0183] In the formula: c represents the area where power is restored; C represents the set of radial topologies of the guaranteed network;

[0184] (5) Node power balance constraints:

[0185]

[0186] In the above formula: within the isolated island k' and These represent the active power and reactive power flowing from node i' to node g' on line (i', g'), respectively. and This represents the active and reactive power of the load connected to node g'; and The active and reactive power injected into the distributed power sources on node g'; r i'g' and x i'g' These are the resistance and reactance on the line (i', g'), respectively; Let be the square of the current amplitude on line (i', g').

[0187] To verify the effectiveness of this invention, the proposed method was simulated and analyzed using an IEEE 33-node system. The system diagram is shown below. Figure 4 As shown, the optical storage system is connected at system nodes 3, 6, 13, and 30 respectively. Figure 4 ES1 to ES4 represent energy storage devices; DG1 to DG4 represent distributed power sources.

[0188] Figure 5 and Figure 6 Simulation results of node voltage and load recovery rate are presented, comparing the results with those of a distribution network recovery strategy that does not consider the controllable output of photovoltaic power generation (comparison strategy). Figure 5 and Figure 6 It can be seen that the method of this invention can smooth out the inherent volatility of photovoltaic power, stabilize photovoltaic output, and meet the requirements of power system restoration for stable power output. The stable photovoltaic output obtained by the method of this invention can effectively reduce system voltage fluctuations and make the system operation more stable. Moreover, the load recovery rate of the method of this invention can reach 96.98%, and it can maintain normal power supply to the restored load for at least 148 minutes. Compared with the comparative strategy, the reliability of power supply restoration of the distribution network that does not consider the controllable output of photovoltaic power generation is reduced, and the power supply time decreases more rapidly.

[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for restoring distribution network output based on the controllability of photovoltaic power generation, characterized in that, include: Step 1: Obtain meteorological information and photovoltaic equipment information for the forecast date, and use a photovoltaic output forecasting method based on multi-factor correction to forecast the photovoltaic output; Step 2: Generate a photovoltaic output prediction curve based on the photovoltaic output prediction results, decompose the photovoltaic output prediction curve, and decompose the photovoltaic power station into multiple virtual distributed power combinations that are available for specific time periods and have controllable output power. Step 3: Based on the distribution network data, selectively restore unrestored loads using constraints, construct an objective function and solve it to obtain the islanding results and load restoration order. The distribution network data includes the virtual distributed power source data. Step 4: Based on the islanding results and load restoration sequence, restore the distribution network; The objective function in step 3 is expressed by the formula: In the formula: Represents the total set of recovery phases; Represents the set of nodes in a distribution network; Indicates the duration of the recovery phase; Indicates an island formed after a failure. internal nodes Active power under load; Represents the node weight coefficient; Let be a binary variable, where Representative node The load awaits power restoration. Representative node Power has been restored to the load; This represents a collection of photovoltaic power generation devices and energy storage devices; This represents the network loss of the entire system in stage t. represent Phased photovoltaic power generation devices Output power; represent Phased photovoltaic power generation devices Output power; These are the weight coefficients of the objective function; This represents a collection of isolated islands.

2. The distribution network restoration method based on the controllability of photovoltaic power generation to restore output according to claim 1, characterized in that, Step 1 includes: Establish a physical model for photovoltaic output, obtain meteorological and equipment information for the forecast date, and determine the preliminary photovoltaic power forecast for the forecast date; The factors influencing photovoltaic power output prediction were identified, and the importance weight vector of the influencing factors corresponding to each variable importance assessment method was obtained by using multiple variable importance assessment methods. The fuzzy multi-criteria method is used to evaluate the various variable importance assessment methods, and the optimal assessment method is determined from among the various variable importance assessment methods. The correction coefficient for photovoltaic output is determined based on the weight vector of the importance of the influencing factors corresponding to the optimal evaluation method. The final photovoltaic power forecast is determined based on the correction factor and the preliminary photovoltaic power forecast.

3. The distribution network restoration method based on the controllability of photovoltaic power generation to restore output according to claim 2, characterized in that, The various variable importance assessment methods include at least two of the following: a variable importance assessment method based on the RF algorithm and Gini coefficient, a variable importance assessment method based on the reduction of average precision, and a variable importance assessment method based on the entropy weight method.

4. The distribution network restoration method based on the controllability of photovoltaic power generation to restore output according to claim 2, characterized in that, The fuzzy multi-criteria method is used to evaluate the importance assessment methods for the various variables, including: A weight matrix is ​​constructed using the determined influencing factors for photovoltaic power output prediction and the weights of the influencing factors corresponding to the importance assessment methods for each variable. ; According to the weight matrix The optimal evaluation method can be expressed by the following formula: ； in, ， ， ； ； ； in, Indicates the number of influencing factors; Represents the decision vector; , and These represent the decision values ​​corresponding to each variable importance assessment method, and the final determined optimal assessment method is... , and The variable importance assessment method corresponding to the maximum value in the value is... This indicates the number of methods used in the fuzzy multi-criteria method for evaluating variable importance. Represents the weight vector; A parameter indicating the number of variables involved in the decision-making process; Indicates by the first The importance assessment method for the first variable yields the following results for the first variable. The importance values ​​of each influencing factor.

5. A distribution network restoration method based on the controllability of photovoltaic power generation to restore output, as described in claim 2, characterized in that, Based on the importance weight vector of influencing factors corresponding to the optimal evaluation method, the correction coefficient for photovoltaic output is determined, including: Assume the weight vector of the importance of influencing factors corresponding to the optimal evaluation method. The predicted value for each influencing factor is obtained. ,but; Based on the predicted value for each influencing factor and weight vector The comprehensive prediction score S is obtained; The correction coefficient is determined based on the comprehensive prediction score.

6. The distribution network restoration method based on the controllability of photovoltaic power generation to restore output according to claim 5, characterized in that, Based on the predicted value for each influencing factor and weight vector The comprehensive prediction score S is obtained, which can be expressed by the formula: ， in, , Indicates the first Predicted values ​​for each influencing factor; , This indicates that the optimal evaluation method is used to obtain the result for the th... The importance weights of each influencing factor This indicates the number of influencing factors.

7. A distribution network restoration method based on the controllability of photovoltaic power generation to restore output, as described in claim 6, is characterized in that... The correction coefficient is determined based on the comprehensive prediction score, expressed by the formula: ， in: ； ； and This indicates the parameters that are set. and These represent the upper and lower limits of the value range for the influencing factors.

8. The distribution network restoration method based on the controllability of photovoltaic power generation to restore output according to claim 1, characterized in that, Step 2, which involves decomposing the photovoltaic power output prediction curve, includes: Step 31: Extract the node data of the photovoltaic power output prediction curve and calculate the variance of the node data. The node data represents the photovoltaic output power prediction results obtained at each time point; Step 32: Set the constant sequence Among them, let ,right After rounding to find even numbers, we get Using sequences The horizontal axis of the photovoltaic power output prediction curve is divided into partitions; Step 33: Set the point with the minimum slope between two points in each interval. This is the decomposition point; Step 34: and , and and Decomposition points are paired up to create decomposition pairs; Step 35: Perform rectangular decomposition on the decomposed pairs to obtain the photovoltaic output virtual distributed power source.

9. The distribution network restoration method based on the controllability of photovoltaic power generation to restore output according to claim 1, characterized in that, The constraint functions in step 3 include: (1) Capacity constraint: In the formula, Indicates an isolated island The remaining capacity of internal photovoltaic power generation and energy storage devices; (2) Energy storage constraints: In the formula, and They represent in Moments, the island Energy storage devices on internal nodes Remaining generating capacity and discharging power; , , and They represent isolated islands. The maximum and minimum capacity, maximum discharge power, and minimum discharge power of the internal energy storage device; G represents the collection of energy storage devices; (3) Node voltage constraints: In the formula: Represents a node voltage amplitude The lower limit; Represents a node The upper limit of voltage amplitude; (4) Network radial constraints: In the formula: Indicates the area where power has been restored; This represents a set of networks with radial topologies. (5) Node power balance constraints: In the formula: isolated island Inside and They represent the lines respectively Above, from the node Outflow, flow to node Active power and reactive power; and Represents a node The active and reactive power of the load connected to the top; and For nodes The active and reactive power injected by the distributed power sources on the platform; and The lines are respectively Resistance and reactance; For the line The square of the amplitude of the current.

Images