A power supply unit division method considering photovoltaic and load uncertainty

Abstract

The application discloses a power supply unit division method considering photovoltaic and load uncertainty, and comprises the following steps: modeling the uncertainty of photovoltaic output and load based on a K-means clustering multi-scene method; constructing a power supply unit division model considering photovoltaic output and load uncertainty, with the least number of power supply unit divisions and the maximum load rate balance of transfer lines in each power supply unit as the target; and solving the power supply unit division model considering photovoltaic output and load uncertainty, so as to obtain the power supply unit division result. The application makes up for the deficiency of the traditional power supply unit division method under the background of large-scale access of distributed photovoltaic, and provides a theoretical basis for the construction of a new type of power distribution system.

Description

Technical Field

[0001] Based on the traditional distribution network power supply unit division method, this patent proposes a power supply unit division method that considers the uncertainty characteristics of photovoltaic and load, and takes into account the probability information of photovoltaic and load. It belongs to the field of distribution network grid division. Background Technology

[0002] As a crucial component of distribution network grid planning and the construction of new distribution systems, the method for dividing distribution network power supply units needs to closely align with current development trends. However, due to the intermittent and fluctuating nature of distributed photovoltaic (PV) power output, its large-scale integration will significantly impact traditional distribution network power supply unit division methods. Furthermore, the stochastic characteristics of power load, influenced by factors such as weather and user behavior, may also lead to variations in the power supply unit division results. To fully implement the new development philosophy, it is necessary to propose a power supply unit division method that considers the uncertainties of both PV and load in the context of large-scale distributed PV integration, thereby supporting the construction of new distribution systems. Summary of the Invention

[0003] To address the shortcomings and deficiencies in existing technologies, this patent proposes a power supply unit partitioning method that considers the uncertainty of photovoltaic and load conditions. This method compensates for the deficiencies of traditional power supply unit partitioning methods in the context of large-scale distributed photovoltaic access, and provides a theoretical basis for the construction of new power distribution systems.

[0004] Specifically, the power supply unit partitioning method considering photovoltaic and load uncertainties proposed in this application includes:

[0005] (1) A multi-scenario method for modeling the uncertainty of photovoltaic output and load based on K-means clustering;

[0006] (2) A method for constructing a power supply unit division model that considers the uncertainty of photovoltaic output and load, with the goal of minimizing the number of power supply units and balancing the maximum load rate of the transfer lines within each power supply unit;

[0007] (3) Solution method for power supply unit division model considering the uncertainty of photovoltaic output and load, including five parts: sequential division of inter-station and intra-station power supply units, calculation method of net load considering DG output, division process of inter-station power supply units, division process of intra-station power supply units, and acquisition of final division scheme.

[0008] Step (1), based on the K-means clustering multi-scenario method for uncertainty modeling of photovoltaic output and load, specifically includes:

[0009] Scenario analysis is a commonly used and effective method for handling uncertain problems. Essentially, it discretizes the stochastic problem into multiple scenarios, thereby transforming the uncertain mathematical problem across all scenarios into a definite mathematical problem within a single scenario for solution, avoiding the need to build complex and difficult-to-solve stochastic mathematical models. However, this approach suffers from problems such as large data scale and long computation time.

[0010] In research, when dealing with large datasets, data clustering is often used. Data clustering is used to group a given dataset, dividing a large dataset into smaller, more manageable datasets based on the inherent similarity and correlation between data points; these are typically called "classes." Objects within the same class are highly similar, while objects in different classes are very similar.

[0011] K-means clustering is a type of partitioning method, and it is a typical distance-based clustering method. K-means clustering uses Euclidean distance as a similarity evaluation criterion, assuming that the distance between objects in the same cluster should be as small as possible, and the distance between objects in different clusters should be as large as possible.

[0012] Let the sample dataset D = {X1, X2, ..., X} n}, where n represents the number of samples, and the i-th sample X i ={x i1 ,x i2 ,...,x ip}, where p represents the number of attributes or types of indicators in a sample, and K represents the number of clusters. In K-means clustering, any two samples X... i With X f The Euclidean distance between them is as follows:

[0013]

[0014] At the beginning of clustering, K samples are randomly selected as initial cluster centers. Then, according to the above formula, samples that are not cluster centers are classified into the category C represented by the cluster center h that is closest to them. h middle.

[0015] Then the cluster centers are updated using the following formula:

[0016]

[0017] In the formula, φ(X) i ) represents the h-th type C h The samples included, |C h | represents the h-th type C h The number of samples included, m h Represents the h-th type Ch The mean of the sample.

[0018] The goal of K-means clustering is to minimize the squared error criterion function, which has the following form:

[0019]

[0020] The cluster centers m are continuously updated using the above formula. h And the samples φ(X) contained in various classes i This continues until the squared error criterion function value no longer changes or the set number of iterations is reached.

[0021] After the iteration is complete, read the cluster centers m at this point. h This is the required typical sample, φ(X) i This refers to the class corresponding to each typical sample.

[0022] This patent employs a multi-scenario method to address the uncertainties in DG output and load. By collecting the output and load of DG at each moment across all scenarios and processing them using K-means clustering, the typical output of DG at each moment and its probability of occurrence, as well as the load timing characteristics and their probability of occurrence, are obtained. The DG output characteristics are combined with the user-side load characteristics at the DG access location to obtain the net load scenario of the user-side load. Based on scenario modeling according to equation (4), the distribution of all scenarios of the user-side net load at each moment within the power supply unit is obtained.

[0023]

[0024] Where A is the distribution of all scenarios at each time step; N is the number of scenarios; and τ is the net load value for each scenario at each time step.

[0025] Step (2), which considers the uncertainty of photovoltaic output and load, specifically includes the following:

[0026] 1) Establish the objective function for the power supply unit partitioning model that considers the uncertainties of photovoltaic output and load, specifically:

[0027] The division of power supply units in medium-voltage distribution networks should aim to minimize the number of power supply units. This involves improving the utilization efficiency of each feeder and reducing the number of outgoing lines from substations by exploring the load characteristic matching between feeders. However, simply considering the minimum total number of power supply units has a step-like effect, easily generating multiple planning schemes with the same number of power supply units. Therefore, in addition to minimizing the number, it is also necessary to ensure the balance of the maximum load rate of the transfer lines within each power supply unit before further evaluating the merits of the power supply unit division. Considering the balance of the maximum load rate of transfer lines on a unit-by-unit basis also helps reduce network losses in distribution network planning. The specific objective function is shown in the following formula:

[0028] minY

[0029] Y = Y zj +Y zn

[0030]

[0031] in, γ represents the average of the expected maximum maximum load rate of the transfer lines within each power supply unit; imax Y represents the maximum expected maximum load rate of the transfer lines in the i-th power supply unit; β is the variance of the expected maximum load rate of the transfer lines in each power supply unit, which describes the balance among the maximum load rates of the transfer lines in each power supply unit; Y, Y zj and Y zn These represent the total number of power supply units, the number of inter-station power supply units, and the number of intra-station power supply units, respectively.

[0032] 2) Establish constraints for the power supply unit partitioning model that considers the uncertainties in photovoltaic output and load, specifically:

[0033] (a) Maximum load rate constraint of the transfer lines within the power supply unit

[0034] Assuming the average net load is maximized in the l-th power supply unit at time t1, and the maximum load rate constraint of the transfer lines within the power supply unit is based on time t1, Q l t1 The net load of the l-th power supply unit at time t1 is uncertain, and the constraints are as follows:

[0035]

[0036] Where cosφ is the power factor; L max This represents the maximum transmission capacity of the line.

[0037] Chance-constrained programming can transform the formula into:

[0038]

[0039] Where ε is the confidence level for the constraint to hold, the above equation can be transformed to obtain:

[0040]

[0041] The uncertainty is handled by using a multi-scenario method. The distribution of the net load of the power supply unit at each moment across all scenarios is obtained by the following formula.

[0042]

[0043] in, represents the net load value of the l-th power supply unit at each moment in each scenario; N is the number of scenarios.

[0044] (b) Feeder load factor constraint

[0045] Assuming the average net load on feeder i is maximized at time t2, the feeder load factor constraint should be based on time t2, P i t2 Let be the net load of feeder i at time t2.

[0046] The load factor constraint for the i-th feeder is shown in the following formula:

[0047]

[0048] Transforming the above equation using chance constraints yields the following result:

[0049]

[0050] The uncertainty is handled by using a multi-scenario method. The distribution of the feeder's net load across all scenarios at each moment is obtained by the following formula.

[0051]

[0052] in, is the net load value of feeder i at each time in each scenario; N is the number of scenarios.

[0053] The solution method for the power supply unit partitioning model considering the uncertainty of photovoltaic output and load in step (3) specifically includes:

[0054] 1) Sequential division of power supply units between and within stations

[0055] To determine the power supply unit division for load characteristic matching between interconnecting feeders, the first step is to divide the inter-station power supply units based on feeder load rate constraints, fully utilizing the power supply capacity between substations. Under the corresponding load rate constraints, N is generated through multiple iterations by changing the positions of the inter-station feeders. step A scheme for dividing power supply units between stations is developed. The average load simultaneity rate within each inter-station power supply unit is used as an indicator to evaluate the merits of the division, and multiple inter-station power supply unit division schemes are selected. Based on this, an intra-station power supply unit division is carried out considering the load characteristic matching between interconnecting feeders, forming a power supply unit division scheme under the constraint of inter-station feeder load rate, thereby achieving the sequential division of inter-station and intra-station power supply units.

[0056] 2) Calculation method for net load considering DG output

[0057] The average value of the net load characteristic curve considering the output of DG is taken to obtain the net load considering the output of DG.

[0058]

[0059] in, The net load time-series output is the average; N is the number of scenarios; μ i (t) represents the net load time-series output of the i-th scenario.

[0060] 3) Inter-station power supply unit division process

[0061] The division of inter-station power supply units first determines the number of feeder lines between substations based on the load scale and power supply model requirements within the power supply grid. This division is then carried out through feeder load rate control to explore the power supply capacity between substations. The specific division process is illustrated using a set of inter-station power supply units from substations S1 and S2 as an example:

[0062] Step 1: Based on the geographical proximity of each substation, determine the furthest distance between any two substations forming an inter-station connection. Assume feeder a... i With b j This is the initial position of the feeder between the two substations. Let k = 1, and set the load rate η of the feeder between the substations.

[0063] Step 2: Initially set i=1, j=1, calculate the distance from each load point in substations S1 and S2 to the feeder a. i With feeder b j The weighted distance is used to assign the load with the smallest weighted distance to feeder a. i With b j The power supply range.

[0064] The specific expressions for the weighted distance and weighting factor are as follows:

[0065] d m,i =d' m,i ω1ω2

[0066] d n,j =d' n,j ω1ω2

[0067]

[0068]

[0069] Among them, inter-station feeder a i With b j This forms a power supply unit, where ω1 represents the weighting factor, which is related to the peak-valley difference rate of the power supply unit, and q is the amplification coefficient of the weighting factor; η max Indicates the load point and the photovoltaic feeder a i or feeder bj Then, the maximum value of the net load time-sequential output within the power supply unit, η min This represents the minimum net load time-series output average within the power supply unit after the addition of load or photovoltaic power. When the addition of load or photovoltaic power helps improve the peak-valley difference of the power supply unit, the weighting factor ω1 value will decrease, and the weighting distance will also decrease, increasing the probability that the load belongs to the power supply unit, thereby ensuring the matching of source and load characteristics within the power supply unit. Indicates the load point and the photovoltaic feeder a i or feeder b j Then, the variance of the net load within the power supply unit at time t is used to describe the degree of fluctuation of the net load within the power supply unit. When the addition of a load point or photovoltaic (PV) reduces the variance of the net load, it indicates that the addition of the load reduces the degree of fluctuation and uncertainty of the net load. The weighting factor ω2 will decrease, and the weighted distance will decrease accordingly, which is beneficial for the addition of a load point or PV to the power supply unit. m,i 'and d m,i These represent the m-th load to feeder a, respectively. i Euclidean distance and weighted distance; d n,j 'and d n,j These represent the nth load to feeder b. j Euclidean distance and weighted distance.

[0070] Step 3: Connect feeder a i With b j The angle is updated according to the location of the load, so that the feeder passes through the geographical center of the load.

[0071] Step 4: Save feeder a i With b j The power supply range division scheme.

[0072] Step 5: Calculate feeder a i With b j Probabilistic information on net load factor.

[0073] Step 6: Determine feeder a i Is the load rate at a certain confidence level less than the preset load rate constraint? If feeder a i If the maximum load rate is less than the preset load rate constraint at a certain confidence level, let i = i + 1, and jump to step 2 to continue dividing feeder a. i If the power supply range is within the specified range, otherwise continue to step 7.

[0074] Step 7: Output feeder a i-1 Scheme for dividing the power supply area.

[0075] Step 8: Determine feeder b jIs the maximum load rate under a certain confidence level less than the preset load rate constraint? If feeder b j If the maximum load rate under a certain confidence level is less than the preset load rate constraint, let j = j + 1, and jump to step 2 to continue dividing feeder b. j If the power supply range is within the specified range, otherwise continue to step 9.

[0076] Step 9: Output feeder b j-1 Scheme for dividing the power supply area.

[0077] Step 10: Determine if the iteration count N has been reached. step If yes, proceed to step 11; otherwise, switch to feeder a. i With b j The initial angle is rotated by △θ1 (△θ1 is the set feeder adjustment angle, set to 3°), and then the process returns to step 2.

[0078] Step 11: Calculate the maximum load rate of the transfer lines in each power supply unit division scheme under a certain confidence level.

[0079] Step 12: Output a power supply unit division scheme that satisfies the maximum load rate constraint of the transfer lines within the power supply unit, and optimize the inter-station power supply unit division scheme based on the minimum average load simultaneity rate within each inter-station power supply unit.

[0080] Step 13: Output multiple sets of inter-station power supply unit division schemes.

[0081] 4) Substation power supply unit division process

[0082] The division of power supply units within a substation is based on the division scheme of power supply units between substations. Within the power supply area of ​​each substation, a group of interconnected feeder groups within the substation is used as the research unit to divide the power supply units considering the load characteristics matching between the interconnected feeders.

[0083] First, taking the substation as the center, several center lines are generated within the substation area according to the principle of equal angle division. Every two adjacent center lines are interconnected to form a substation power supply unit. Second, the weighted distance from each load point (including photovoltaic) to each center line is calculated. The definition of weighted distance is the same as that in the division of inter-station power supply units. According to the principle of minimizing weighted distance and satisfying the chance constraint of the maximum load rate of the transfer line within each power supply unit, each load point (including photovoltaic) is divided into the power supply range of each center line. This ensures that the load characteristics of the substation power supply units are matched. After the power supply range of each center line is divided, its angle is updated according to the location of the load carried by the center line so that the center line passes through the center of the geographical location of the load it carries. Finally, multiple iterations are performed to complete the division of the substation power supply units.

[0084] 5) Obtaining the final partitioning scheme

[0085] Iterate through all feasible schemes and compare the number of power supply units and the variance of the maximum load rate of the transfer lines within each power supply unit. Based on the minimum number of power supply units, select the power supply unit partitioning scheme with the smallest variance of the maximum load rate of the transfer lines within each power supply unit.

[0086] Beneficial effects:

[0087] A method for dividing power supply units that considers the uncertainties of photovoltaics and loads is proposed, which can properly solve the problem of multiple uncertainties between sources and loads, make up for the shortcomings of traditional power supply unit division methods in the context of large-scale distributed photovoltaic access, and provide a theoretical basis for the construction of new power distribution systems. Attached Figure Description

[0088] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0089] Figure 1 This is a flowchart illustrating the power supply unit partitioning based on photovoltaic and load uncertainties in this invention.

[0090] Figure 2 This is a geographic information map of the area to be divided in this implementation case.

[0091] Figure 3 This diagram shows the power supply unit partitioning results considering photovoltaic and load uncertainties in this implementation case. Detailed Implementation

[0092] To make the structure and advantages of the present invention clearer, the structure of the present invention will be further described below with reference to the accompanying drawings.

[0093] Combination Figure 1 The present invention provides a detailed explanation of the power supply unit partitioning process that considers photovoltaic and load uncertainties. The specific steps are as follows:

[0094] Step 1: Uncertainty modeling method for photovoltaic power output and load based on K-means clustering multi-scenario method;

[0095] Step 2: A method for constructing a power supply unit partitioning model that considers the uncertainty of photovoltaic output and load, with the goal of minimizing the number of power supply units and balancing the maximum load rate of the transfer lines within each power supply unit;

[0096] Step 3: Solution method for power supply unit partitioning model considering photovoltaic output and load uncertainty, including five parts: sequential partitioning of inter-station and intra-station power supply units, calculation method of net load considering DG output, partitioning process of inter-station power supply units, partitioning process of intra-station power supply units, and acquisition of final partitioning scheme.

[0097] The planned area is a power grid, as shown in the attached diagram. Figure 2 As shown, this includes three 2×63MVA substations, each with 160, 156, and 150 loads respectively. The load types and locations are shown in the diagram below. The 10kV main line is model LGJ-185, with a feeder capacity of 7.8MW. The loads are categorized into four types: residential, commercial, industrial, and administrative. The power supply area of ​​the three substations covers 223 street nodes and 407 street sections. The geographical locations of each street node, street section, and load are shown in the diagram below. Figure 2 As shown, the dashed lines represent the boundaries of the power supply range of each substation in the power supply grid, and the red nodes indicate load points configured with distributed generation. Considering the uncertainty of load size and distributed generation output, the confidence level ε is 0.95. The upper limit of the slack load rate of inter-station feeders is 58%, and the slack load rate step size is 1%. The power supply units are divided using a single-tie wiring mode as an example. The division results obtained according to the method proposed in the patent are as follows. Figure 3 As shown.

Claims

1. A method for dividing power supply units considering photovoltaic and load uncertainties, characterized in that, The method includes: Step 1: Uncertainty modeling method for photovoltaic power output and load based on K-means clustering multi-scenario method; Step 2: A method for constructing a power supply unit partitioning model that considers the uncertainty of photovoltaic output and load, with the goal of minimizing the number of power supply units and balancing the maximum load rate of the transfer lines within each power supply unit; Step 3: Solution method for power supply unit partitioning model considering photovoltaic output and load uncertainty, including five parts: sequential partitioning framework of inter-station and intra-station power supply units, calculation method of net load considering DG output, partitioning process of inter-station power supply units, partitioning process of intra-station power supply units, and acquisition of final partitioning scheme. (1) Sequential division of power supply units between and within stations To determine the power supply unit division for load characteristic matching between interconnecting feeders, the first step is to divide the inter-station power supply units based on feeder load rate constraints. This fully leverages the power supply capacity between substations. Under the corresponding load rate constraints, the division is iterated multiple times by changing the positions of the inter-station feeders to generate the necessary units. N step The inter-station power supply unit division scheme is developed, and the average load simultaneity rate within each inter-station power supply unit is used as an indicator to evaluate the merits of the division. Multiple inter-station power supply unit division schemes are selected, and based on these, intra-station power supply unit division is carried out considering the load characteristic matching between interconnecting feeders. This results in a power supply unit division scheme under the constraint of inter-station feeder load rate, thereby achieving the sequential division of inter-station and intra-station power supply units. (2) Calculation method for net load considering DG output The average value of the net load characteristic curve considering DG output is taken to obtain the net load amount considering DG output. ,in, This represents the average output during the net load sequence. N The number of scenes; μ i ( t ) is the first i Net load time-series output for each scenario (3) Inter-station power supply unit division process The division of inter-station power supply units first determines the number of outgoing lines between substations based on the load scale and power supply model requirements in the power supply grid. This division is then carried out through load rate control of the inter-station feeders to explore the power supply capacity between substations. The specific division process is illustrated using a set of inter-station power supply units from substations S1 and S2 as an example: Step 1: Based on the geographical proximity of each substation, determine the furthest distance between any two substations forming an inter-station connection, assuming a feeder... a i and b j It is the initial position of the feeder between the two substations, making k =1, setting the load rate of the inter-station feeder. , Step 2: Initial Command i =1, j =1, calculate the feeder from each load point in substations S1 and S2. a i With feeder b j The weighted distance is used to assign the load with the smallest weighted distance to the feeder. a i and b j Power supply range, The specific expressions for the weighted distance and weighting factor are as follows: ， ， ， Among them, inter-station feeders a i and b j To form a power supply unit ω 1 represents the weighting factor, which is related to the peak-valley difference rate of the power supply unit. q This is the amplification factor for the weighting factor; η max Indicates the load point and the photovoltaic feeder a i or feeder b j Then, the maximum value of the net load time-sequential output within the power supply unit. η min This represents the minimum net load time-series output of the power supply unit after the addition of load or photovoltaic power. The weighting factor is used when the addition of load or photovoltaic power helps improve the peak-valley difference of the power supply unit. ω A smaller value (1) results in a smaller weighted distance, increasing the likelihood that a load point belongs to that power supply unit, thereby ensuring source-load characteristic matching within the power supply unit. Indicates the load point and the photovoltaic feeder a i or feeder b j After that, the net load within the power supply unit was in the [number]th [year]. t The variance of the net load size at any given time describes the degree of fluctuation in the net load size within a power supply unit. When the addition of a load point and photovoltaic power reduces the variance of the net load size, it indicates that the addition of this load reduces the degree of fluctuation and uncertainty in the net load size. (Weighting factor) ω The value of 2 will decrease, and the weighted distance will decrease accordingly, which is beneficial for load points or photovoltaic systems to be added to the power supply unit. d m,i ’ and d m,i They represent the first m Load to feeder a i Euclidean distance and weighted distance; d n,j ’ and d n,j They represent the first n Load to feeder b j Euclidean distance and weighted distance, Step 3: Connect the feeder a i and b j The angle is updated according to the location of the load, so that the feeder passes through the geographical center of the load. Step 4: Save the feeder a i and b j Power supply range division scheme, Step 5: Calculate the feeder a i and b j Probabilistic information on net load factor. Step 6: Determine the feeder a i Is the load rate at a certain confidence level less than the preset load rate constraint? If the feeder... a i Under a certain confidence level, the maximum load rate is less than the preset load rate constraint, so that... i = i +1, jump to step 2, continue dividing the feeder. a i If the power supply range is within the specified range, otherwise continue to step 7. Step 7: Output feeder a i-1 Power supply area division scheme, Step 8: Determine the feeder b j Is the maximum load rate under a certain confidence level less than the preset load rate constraint? If the feeder... b j If the maximum load rate under a certain confidence level is less than the preset load rate constraint, then... j = j +1, jump to step 2, continue dividing the feeder. b j If the power supply range is within the specified range, otherwise continue to step 9. Step 9: Output Feeder b j-1 Power supply area division scheme, Step 10: Determine if the number of iterations has been reached. N step If yes, proceed to step 11; otherwise, turn the feeder... a i and b j Initial angle rotation △ θ 1 Then return to step 2. Step 11: Calculate the maximum load rate of the transfer lines in each power supply unit division scheme under a certain confidence level. Step 12: Output a power supply unit partitioning scheme that satisfies the maximum load rate constraint of the transfer lines within the power supply unit, and optimize the inter-station power supply unit partitioning scheme based on the minimum average load simultaneity rate within each inter-station power supply unit. Step 13: Output multiple sets of inter-station power supply unit division schemes.

2. The method for dividing power supply units considering photovoltaic and load uncertainties as described in claim 1, characterized in that, The uncertainty modeling method for photovoltaic power output and load based on K-means clustering multi-scenario method specifically includes: A multi-scenario approach is employed to address the uncertainties in distributed generation (DG) output and load. This involves collecting the output and load data of DG at every moment across all scenarios and processing them using K-means clustering. This yields the typical output of DG at each moment, its probability of occurrence, and the timing characteristics and probabilities of load occurrence. Combining the DG output characteristics with the user-side load characteristics at the DG connection location, the net load scenarios for the user side are obtained. Based on the scenario modeling formula below, the distribution of the net user-side load for all scenarios at each moment within the power supply unit is obtained. ,in, A The distribution of all scenes at each moment; N The number of scenes; τ Net load values ​​for each scenario at each time point.

3. The method for dividing power supply units considering photovoltaic and load uncertainties as described in claim 1, characterized in that, The power supply unit partitioning model construction method specifically includes: The objective function for the power supply unit partitioning model considering photovoltaic output and load uncertainties is established as follows: In addition to minimizing the number of power supply units, it is also necessary to ensure that the maximum load rate of the transfer lines within each power supply unit is balanced before further evaluation of the merits of the power supply unit division. The specific objective function is shown in the following formula: ， ， ， in, This represents the average of the maximum expected maximum load rate of the transfer lines within each power supply unit; γi max represents the first i The maximum expected maximum load rate of the transfer lines in each power supply unit; β The variance of the expected maximum load rate of the transfer lines within each power supply unit is used to describe the balance among the maximum load rates of the transfer lines within each power supply unit. Y , Y zj and Y zn These represent the total number of power supply units, the number of inter-station power supply units, and the number of intra-station power supply units, respectively. The constraints for establishing a power supply unit partitioning model that considers uncertainties in photovoltaic output and load are as follows: (1) Maximum load rate constraint of transfer lines within the power supply unit Assuming in t 1st moment l The average net load size is the largest among the power supply units, and the maximum load rate constraint of the transfer lines within the power supply unit is based on... t Using time 1 as the standard, Q l t1 For the first l Each power supply unit t The magnitude of the net load at time 1 is uncertain, and the constraints are as follows: ， Among them, cos φ Power factor; L max This represents the maximum transmission capacity of the line. Chance-constrained programming can transform the formula into: ,in, ε To determine the confidence level at which the constraints hold, we can transform the above equation to obtain: To handle uncertainties, a multi-scenario approach is adopted. The distribution of the net load of the power supply unit at each moment across all scenarios is obtained by the following formula: ,in, For the first l Net load values ​​for each power supply unit at each moment in each scenario; N For the number of scenes, (2) Feeder load factor constraint Assuming in t 2-hour feeder i The average net load is the largest, and the feeder load rate constraint should be based on... t Using time 2 as the standard, P i t2 For feeder i The load is t The size of the net load at time 2, No. i The load factor constraint for each feeder is shown in the following formula: ， Transforming the above equation using chance constraints yields the following result: ， To handle uncertainty, a multi-scenario approach is used. The distribution of the feeder's net load across all scenarios at each moment is obtained by the following formula: ， in, For feeder i Net load values ​​for each scenario at each time point; N This represents the number of scenes.

4. The method for dividing power supply units considering photovoltaic and load uncertainties as described in claim 1, characterized in that, The aforementioned method for solving the power supply unit partitioning model considering photovoltaic output and load uncertainties also includes: (4) Process of dividing the power supply units within the station The division of power supply units within a substation is based on the inter-substation power supply unit division scheme. Within the power supply area of ​​each substation, a group of interconnected feeder groups is used as the research unit to divide the power supply units considering the load characteristics matching between the interconnecting feeders. First, taking the substation as the center, several center lines are generated within the substation area according to the principle of equal angle division. Each pair of adjacent center lines interconnects to form a substation power supply unit. Second, the weighted distance from each load point to each center line is calculated. The definition of weighted distance is the same as that used in the division of inter-substation power supply units. Following the principle of minimizing weighted distance and satisfying the chance constraint of the maximum load rate of the transfer lines within each power supply unit, each load point is assigned to the power supply range of each center line. This ensures that the load characteristics of the substation power supply units are matched. After the power supply range of each center line is divided, its angle is updated according to the location of the load it carries, ensuring that the center line passes through the center of the geographical location of the load it carries. Finally, multiple iterations are performed to complete the division of the substation power supply units. (5) Obtaining the final partitioning scheme Iterate through all feasible schemes, compare the number of power supply units in each scheme with the variance of the maximum load rate of the transfer lines within each power supply unit, and select the power supply unit partitioning scheme with the smallest number of power supply units and the smallest variance of the maximum load rate of the transfer lines within each power supply unit.

Images