A source-load collaborative planning method for active distribution network considering optimal power flow

Abstract

The application belongs to the technical field of electric power, and provides a source-load collaborative planning method of active distribution network considering optimal power flow, wherein, first, data is acquired from a monitoring system of the distribution network; then, load prediction is performed on typical load historical data based on a prediction combination model of multi-standard nonlinear programming, and then, a source-load collaborative planning objective function and constraint conditions considering optimal power flow are established according to the load prediction data; finally, optimal site selection and distribution ratio of wind power and photovoltaic distributed power connected to the distribution network are multi-objectively optimized through a modified star optimization algorithm; the source-load collaborative planning model considering optimal power flow has an upper planning model taking economy as an objective, and comprehensively considers distributed power investment cost and annual operation reliability cost, and a lower optimal power flow model taking minimization of line loss as an objective, so that the planning result can meet the safe and stable operation requirements, the efficient operation of the power system is realized, and the power supply reliability and power quality are improved.

Description

Technical Field

[0001] This invention belongs to the field of power technology, specifically relating to an active distribution network source-load coordinated planning method that considers optimal power flow. Background Technology

[0002] With the development of power systems, especially the booming new energy industry, renewable energy sources such as wind and solar power have gradually become an important part of the power system. Distributed power sources, as representatives of renewable energy, have become an indispensable part of the power system. Different types of distributed power sources have different generation characteristics and uncertainties. In order to achieve optimal operation of the power system, determining the optimal proportion and combination of new energy sources has become an important research direction.

[0003] This invention explores the optimal site selection and allocation ratio for wind power and distributed photovoltaic (PV) power generation when connected to the distribution network. This optimizes grid resource allocation, ensures grid stability and power supply reliability, and improves the economic efficiency of grid operation. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an active distribution network source-load coordinated planning method that considers optimal power flow, thereby resolving the issues in the prior art. The technical solution adopted by this invention is as follows:

[0005] An active distribution network source-load coordinated planning method considering optimal power flow includes the following steps:

[0006] Step 1: Obtain historical operational data samples of the source payload and process the data;

[0007] Step 2: Perform load forecasting based on typical historical load data using a multi-standard nonlinear programming-based forecasting combination model;

[0008] Step 3: Based on historical data and load forecast results, construct a source-load collaborative planning model that considers optimal power flow;

[0009] Step 4: Solve the source-load collaborative planning model for optimal power flow by improving the planetary optimization algorithm, and realize source-load collaborative planning.

[0010] Furthermore, step 1 includes: obtaining historical load operation data samples from the power distribution network monitoring system, historical load data, weather forecasts, and distributed power generation data; cleaning and normalizing the collected load data to remove outliers and noise.

[0011] Furthermore, step 2 includes:

[0012] Multiple forecasting models are used to predict load data, with each model producing a set of forecast values;

[0013] For each prediction model M i , i∈(1,N); let E=e j The set of errors, j∈(1,L), is used to measure the accuracy of predictions. Errors are measured using various error measurement methods, including mean absolute error, root mean square error, and mean absolute percentage error.

[0014]

[0015] Let be the set of symbols associated with each error, positive for the case where the error is to be minimized, and negative otherwise;

[0016]

[0017] In the formula, Weights associated with each type of error;

[0018] For cases where some error metrics are not differentiable, continuous processing is performed:

[0019]

[0020] Mean absolute error:

[0021]

[0022] Root mean square error:

[0023]

[0024] Average absolute percentage:

[0025]

[0026] T(s i A function used to measure directional changes in a time series, used to determine the consistency of the trend between predicted and actual values:

[0027]

[0028] s i =y i -y i-1

[0029] In the formula, s i =y i -y i-1 This represents the change in the predicted value between time i and i-1;

[0030] T(s t Smooth it into a continuous, differentiable function T′(s) t ):

[0031]

[0032] In the formula, δ m Positive values ​​are used to avoid s i Discontinuity when δ = 0; s It is a positive value, used for numerical stability;

[0033] Smoothed trend accuracy ratio p t And error ratio q′ t calculate:

[0034]

[0035] In the formula, k represents the total number of prediction times;

[0036] The final formula for the smoothed error metric function is as follows:

[0037]

[0038] In the formula, p = p y p z +(1-p y (1-p) z ) represents the probability that the predicted trend is consistent with the actual trend, w = (2p y -1) 2 q z +(2p z -1) 2 q y +4q y q z It is an adjustment term used to correct deviations in directional matching, p y and p z These represent the growth or decline trends of the predicted and actual values, respectively. Represents the variance term;

[0039] The smoothed error metric function is used as the objective function to optimize the prediction model:

[0040] Objective function of the prediction combination model based on multi-criteria nonlinear programming model:

[0041]

[0042] Constraints for load forecasting based on multi-criteria nonlinear programming models:

[0043]

[0044] The model is trained using historical data, and the model parameters are optimized based on error metrics and constraints.

[0045] Furthermore, in step 3, when constructing the source-load collaborative planning model considering optimal power flow, it includes an upper-level objective function, a lower-level objective function, equality constraints, and inequality constraints, wherein:

[0046] Upper-level objective function: Economic efficiency is the objective function.

[0047] min C = C inv +C re

[0048] In the formula, C inv For distributed power investment costs, C re Annual operating reliability cost;

[0049] Lower-level objective function: Minimizing line loss as the objective function.

[0050]

[0051] In the formula, This represents the loss of line l.

[0052] Furthermore, the formula for calculating the investment cost of distributed power sources is as follows:

[0053]

[0054] In the formula, m WT m PV These represent the total number of candidate nodes for wind turbines and solar power, respectively. These are the unit capacity investment costs for wind turbines and solar power, respectively. The rated installed capacity of the fan at node b. The rated installed capacity of the photovoltaic system at node a; CRF is the investment recovery factor;

[0055] The formula for calculating annual operational reliability cost is:

[0056]

[0057] In the formula, E0 is the basic electricity price of the distribution network, and T k，t Let ΔA be the actual power outage time at load point k. k，t ΔB k，t ΔEENS k，t These represent the changes in the average power outage time, average power outage frequency, and expected power shortage value for user k at load point, respectively.

[0058] Furthermore, equality constraints:

[0059]

[0060] In the formula: P Gi Q GiThese are the active and reactive power outputs of node i, P. Di Q Di These are the active and reactive power loads at node i, V. i V j Let θ represent the voltage magnitudes at nodes i and j, where j represents all nodes directly connected to node i. ij G represents the phase angle difference between node i and node j. ij B ij These are the conductance and susceptance of the branch circuit, respectively.

[0061] Furthermore, the inequality constraints include node voltage constraints, branch current constraints, line power flow constraints, installed capacity constraints, and distributed generation penetration constraints, among which:

[0062] Node voltage constraints: The operating voltage of each node is limited to a tolerance of ±5% of the rated voltage.

[0063] 0.95pu≤U i ≤1.05pu

[0064] Where: the operating voltage U of each node i It is limited to a tolerance of ±5% of the rated voltage;

[0065] Branch current constraints:

[0066]

[0067] In the formula: This represents the maximum allowable current for line l;

[0068] Power flow constraints on the line:

[0069]

[0070] In the formula, P ij To save the active power between node i and node j, Q ij To save reactive power between node i and node j, These represent the maximum and minimum apparent power values ​​between power saving i and node j, respectively.

[0071] Installation capacity constraints:

[0072]

[0073] In the formula, P DG，i Q DG，i These represent the active and reactive power of distributed power source access node i, respectively. Let be the maximum active power that node i is allowed to connect to the distributed power source. The maximum reactive power that node i is allowed to connect to a distributed power source;

[0074] Constraints on distributed power penetration:

[0075]

[0076] In the formula, P DG，s For the power output of the distributed power source s, P i P represents the total load requirement of node i. loss This represents the total active power loss of the distribution network.

[0077] Furthermore, step 4 includes:

[0078] Step 4.1: Encode the planets representing the location and capacity of the distributed power source according to the integer encoding rules. Under the constraints, randomly generate the initial population. The initial "sun" is the individual with the largest mass in the population. The smaller the objective function, the larger the mass.

[0079] Step 4.2: Simulate celestial motion using the gravitational mechanism of the planetary optimization algorithm, update the position and mass of each planet, and determine the new "sun" position, which is the global optimal solution in the current population.

[0080] Gravitational torque:

[0081]

[0082] The calculation formulas for each parameter are as follows:

[0083] Planetary mass:

[0084]

[0085] α=f s -f max

[0086] In the formula, a = 2 is a constant, and α = f s -f max The smaller the objective function value of a planet, the greater its mass. i，j f max f s Let represent the objective function values ​​for the i-th or j-th planet, the worst-case planet, and the Sun, respectively.

[0087] The distance between the masses i and j of the two planets:

[0088]

[0089] In the formula, Let Z be the position of planets i and j in each dimension at the t-th iteration; Z is the dimension.

[0090] Step 4.3, Global Search Phase:

[0091]

[0092] In the formula, b is a constant, and rand(0,1) is a random number between 0 and 1. Let be the position of the sun at the t-th time, and β be a coefficient that depends on M. Let be the gravitational value in the t-th iteration. The maximum gravitational value in the t-th iteration;

[0093] A reconnaissance peak mechanism is introduced, using edge solutions as the search area for the bee colony. If a planet is not updated within a certain number of iterations, a new planet position is randomly generated.

[0094] X′ i，j =L d +rand(0,1)×(U d -L d )

[0095] In the formula, U d L d These are the upper and lower limits of the search space, respectively;

[0096] Step 4.4, Local Search Phase:

[0097]

[0098] In the formula, c = 2, T is the maximum number of iterations, t is the current number of iterations, and g is a random number that follows a Gaussian distribution function with mean μ = 0.5 and standard deviation σ = 0.2.

[0099] The present invention has the following beneficial effects:

[0100] This invention proposes a source-load coordinated planning method for active distribution networks that considers optimal power flow. It employs multi-standard nonlinear programming to combine and optimize the prediction model, making the prediction results more closely match the actual load change trend and providing a reliable basis for subsequent planning. The source-load coordinated planning model considering optimal power flow has an upper-level planning model that prioritizes economic efficiency, comprehensively considering the investment cost of distributed power sources and the annual operating reliability cost, while the lower-level optimal power flow model aims to minimize line losses. This ensures that the planning results meet the requirements of safe and stable operation while achieving efficient operation of the power system, improving power supply reliability and power quality. Attached Figure Description

[0101] Figure 1 This is a flowchart of the overall method of the present invention;

[0102] Figure 2 This is a flowchart of the method in step 1 of the present invention;

[0103] Figure 3 This is a flowchart of the method for improving the planetary optimization algorithm in step 4 of the present invention. Detailed Implementation

[0104] The following will be based on embodiments of the present invention. Figures 1-3 The technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0105] This invention first acquires historical load, distributed generation, and distribution network topology data from the distribution network monitoring system. Then, based on a multi-standard nonlinear programming predictive combination model, it performs load forecasting on typical historical load data. Following this, based on the load forecast data, it establishes a source-load collaborative planning objective function and constraints considering optimal power flow. Finally, it uses an improved planetary optimization algorithm to perform multi-objective optimization of the optimal site selection and allocation ratio for wind power and photovoltaic distributed generation access to the distribution network.

[0106] like Figure 1 This invention proposes a source-load coordinated planning method for active distribution networks that considers optimal power flow, specifically including the following steps:

[0107] Step 1: As Figure 2 Obtain historical operational data samples of the source load and process the data:

[0108] Historical load data samples are obtained from the power distribution network monitoring system, historical load data, weather forecasts, and distributed generation data. Data preprocessing is then performed, including cleaning and normalizing the collected load data to remove outliers and noise. This allows the data prediction model to learn from the improved data and generate accurate results and predictions.

[0109] Step 2: Perform load forecasting based on a multi-standard nonlinear programming-based forecasting combination model using typical historical load data:

[0110] First, multiple prediction models, including Long Short-Term Memory Network (LSTM), Convolutional Neural Network (CNN), and Autoregressive Neural Network (ARNN), are used to predict the load data, with each model generating a set of predicted values.

[0111] For each prediction model M i , i∈(1,N); let E=e j Let j∈(1,L) be the set of error metrics for measuring prediction accuracy. Error is measured using error metric methods, including mean absolute error, root mean square error, and mean absolute percentage error.

[0112]

[0113] Let be the set of symbols associated with each error, positive for the case where the error is to be minimized, and negative otherwise.

[0114]

[0115] In the formula, The weights associated with each type of error.

[0116] In particular, some error measurement functions that are not differentiable have been made continuous.

[0117]

[0118] Mean absolute error:

[0119]

[0120] Root mean square error:

[0121]

[0122] Average absolute percentage MAPE:

[0123]

[0124] The error metric function PT-Test, or Pesaran-Timmermann Test, is used as part of the directional error metric to measure the trend consistency between predicted and actual values.

[0125] First, T(s) i ) is a function used to measure directional changes in a time series, mainly used to determine the consistency of the trend between the predicted and actual values.

[0126]

[0127] s i =y i -y i-1

[0128] In the formula, s i =y i -y i-1 This represents the change in the predicted value between time i and i-1.

[0129] Due to the original T(s) t The function is a discrete, discontinuous function, which means it cannot be directly used in the solution process of optimization problems. To optimize in nonlinear programming, the function is smoothed into a continuous, differentiable function T′(s).t ).

[0130]

[0131] In the formula, δ m It is a very small positive value, used to avoid s i Discontinuity when δ = 0. s It is also a very small positive value, used for numerical stability.

[0132] Smoothed trend accuracy ratio p t And error ratio q′ t calculate:

[0133]

[0134] In the formula, k represents the total number of prediction times.

[0135] The final formula for smoothing PT-Test:

[0136]

[0137] In the formula, p = p y p z +(1-p y (1-p) z ) represents the probability that the predicted trend is consistent with the actual trend, w = (2p y -1) 2 q z +(2p z -1) 2 q y +4q y q z It is an adjustment term used to correct deviations in directional matching, p y and p z These represent the growth or decline trends of the predicted and actual values, respectively. This represents the variance term.

[0138] Smoothed error metrics such as PT-Test can be used as objective functions to optimize prediction models.

[0139] Objective function of the prediction combination model based on the multi-criteria nonlinear programming model (MNLP):

[0140]

[0141] Constraints for load forecasting based on the multi-standard nonlinear programming (MNLP) model:

[0142]

[0143] Train the model using historical data, and optimize the model parameters based on error metrics and constraints, such as adjusting the weight parameters in the MNLP model, to ensure that the predictions match reality and improve the accuracy and reliability of the predictions.

[0144] Step 3: Based on historical data and load forecast results, construct a source-load collaborative planning model that considers optimal power flow.

[0145] A source-load collaborative planning model considering optimal power flow is constructed to achieve integrated design of long-term planning and short-term operational optimization of the distribution network. The model includes an upper-level objective function, a lower-level objective function, equality constraints, and inequality constraints. The upper-level objective function aims to optimize the investment and operating costs of distributed generation (DG, such as wind and solar power) from an economic perspective, with its core objective being to minimize the total economic cost of the system. The lower-level objective function focuses on grid performance, with its main objective being to minimize line losses, i.e., optimizing power transmission losses after distributed generation planning to improve the operating efficiency of the power system. Upper-level decisions may influence lower-level behavior, thus partially affecting the achievement of lower-level objectives. However, the upper level cannot completely control the choices made by the lower level; within the limits allowed by the upper-level decisions, the lower level has autonomous decision-making power. Constraints are security constraints on the upper and lower-level objective functions, ensuring that the planning results are both economical and meet the requirements for safe and stable grid operation.

[0146] (1) Upper-level objective function, i.e., planning model: with economy as the objective function:

[0147] min C = C inv +C re

[0148] In the formula, C inv For wind power and distributed photovoltaic (DG) power generation investment costs, C re This represents the annual operational reliability cost. The specific calculation formulas for each cost are as follows:

[0149] Distributed generation (DG) investment costs:

[0150]

[0151] In the formula, m WT m PV These represent the total number of candidate nodes for wind turbines and solar power, respectively. These are the unit capacity investment costs for wind turbines and solar power, respectively. The rated installed capacity of the fan at node b. denoted as Node a, where is the rated installed capacity of the photovoltaic system; CRF is the investment recovery factor.

[0152] Annual operational reliability cost:

[0153]

[0154] In the formula, E0 is the basic electricity price of the distribution network, and T k，t Let ΔA be the actual power outage time at load point k. k，t ΔB k，t ΔEENS k，t These represent the changes in the average power outage time, average power outage frequency, and expected power shortage value for user k at load point, respectively.

[0155] (2) Lower-level objective function, i.e., optimal power flow model: The objective function is to minimize line loss.

[0156]

[0157] In the formula, This represents the loss of line l.

[0158] (3) Equality constraints, i.e. power flow constraints:

[0159]

[0160] In the formula: P Gi Q Gi These are the active and reactive power outputs of node i, P. Di Q Di These are the active and reactive power loads at node i, V. i V j Let θ represent the voltage magnitudes at nodes i and j, where j represents all nodes directly connected to node i. ij G represents the phase angle difference between node i and node j. ij B ij These are the conductance and susceptance of the branch circuit, respectively.

[0161] (4) Inequality constraints:

[0162] Node voltage constraints: The operating voltage of each node is limited to a tolerance of ±5% of the rated voltage.

[0163] 0.95pu≤U i ≤1.05pu

[0164] Where: the operating voltage U of each node i It is limited to a tolerance of ±5% of the rated voltage.

[0165] Branch current constraints:

[0166]

[0167] In the formula: This represents the maximum allowable current for line l.

[0168] Power flow constraints on the line:

[0169]

[0170] In the formula, P ij To save the active power between node i and node j, Q ij To save reactive power between node i and node j, These represent the maximum and minimum apparent power values ​​between power saving node i and node j, respectively.

[0171] Installation capacity constraints:

[0172]

[0173] In the formula, P DG，i Q DG，i These refer to the active and reactive power of distributed generation (DG), specifically, the active and reactive power of the DG connected to node i, which consists of wind and solar power. Let Node i be the maximum active power allowed to connect to the distributed generation (DG). Let be the maximum reactive power that node i is allowed to connect to the distributed generation (DG).

[0174] The constraint on the penetration rate of distributed generation (DG) is to ensure the power output limit of DG, and is subject to the following constraints:

[0175]

[0176] In the formula, P DG，s For the power output of the distributed power source s, P i P represents the total load requirement of node i. loss This represents the total active power loss of the distribution network.

[0177] Step 4: As Figure 3 By improving the planetary optimization algorithm, the source-load collaborative planning model for the optimal power flow in step 3 is solved, thereby realizing source-load collaborative planning.

[0178] Based on historical source-load operation data and load forecasting from step 2, an improved planetary optimization algorithm is used to solve the source-load collaborative planning model, thereby determining the optimal access location and capacity of distributed generation (DG).

[0179] Step 4.1: First, in the search space, the planets representing the locations and capacities of distributed generation (DG) sources (wind and solar) are encoded according to integer encoding rules. Under constraints, an initial population is randomly generated. The initial "sun" is the individual with the largest mass in the population; the smaller the objective function, the larger the mass.

[0180] Step 4.2: Simulate celestial motion using the gravitational mechanism of the planetary optimization algorithm to update the position and mass of each planet. Determine the new "sun" position, which is the global optimum in the current population.

[0181] Gravitational torque:

[0182]

[0183] The specific calculation formulas for each parameter are as follows:

[0184] Planetary mass:

[0185]

[0186] α=f s -f max

[0187] In the formula, a = 2 is a constant, and α = f s -f max The smaller the objective function value of a planet, the greater its mass. i，j f max f s Let represent the objective function values ​​for the i-th or j-th planet, the worst-case planet, and the Sun, respectively.

[0188] The distance between the masses i and j of the two planets:

[0189]

[0190] In the formula, Let Z represent the positions of planets i and j in each dimension at the t-th iteration. Z is the dimension.

[0191] Step 4.3, Global Search Phase: A more refined search is performed in the region close to the global optimum to further improve the solution accuracy.

[0192]

[0193] In the formula, b is a constant, and rand(0,1) is a random number between 0 and 1. Let be the position of the sun at the t-th time, and β be a coefficient that depends on M. Let be the gravitational value in the t-th iteration. This represents the maximum gravitational value in the t-th iteration.

[0194] A reconnaissance peak mechanism is introduced, which uses planets that are far from the "sun" (i.e., edge solutions) as the search area of ​​the swarm. If a planet is not updated in a certain number of iterations, a new planet position is randomly generated.

[0195] X′ i，j =L d+rand(0,1)×(U d -L d )

[0196] In the formula, U d L d These are the upper and lower limits of the search space, respectively.

[0197] Step 4.4, Local Search Phase:

[0198]

[0199] In the formula, c = 2, T is the maximum number of iterations, t is the current number of iterations, and g is a random number that follows a Gaussian distribution function with mean μ = 0.5 and standard deviation σ = 0.2.

[0200] By improving the planetary optimization algorithm, it is possible to simultaneously optimize the location and capacity configuration of distributed power sources while satisfying equality and inequality constraints, thereby achieving the optimal solution for source-load collaborative planning and ensuring a balance between the economy and technical performance of the distribution network.

[0201] This invention utilizes the active power loss sensitivity method to obtain a set of candidate nodes for distributed generation (DG) grid connection. Different capacity combinations of wind turbines and photovoltaics are configured for these candidate nodes. An algorithm is applied to solve the planning model, determining the optimal grid connection location and capacity for wind and photovoltaic distributed generation. The distributed generation (DG) referred to in this invention is a distributed power source composed of wind and photovoltaic power.

[0202] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, alterations, substitutions, or variations made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention shall fall within the protection scope defined by the claims of the present invention.

Claims

1. A source-load coordinated planning method for an active distribution network considering optimal power flow, characterized in that, Includes the following steps: Step 1: Obtain historical operational data samples of the source payload and process the data; Step 2: Perform load forecasting based on typical historical load data using a multi-standard nonlinear programming-based forecasting combination model; Step 3: Based on historical data and load forecast results, construct a source-load collaborative planning model that considers optimal power flow; Step 4: Solve the source-load collaborative planning model considering optimal power flow in Step 3 by improving the planetary optimization algorithm, and realize source-load collaborative planning; Step 2 includes: Multiple forecasting models are used to predict load data, with each model producing a set of forecast values; For each prediction model M i , i∈{1,N}; let E=e j ,j∈{1,L} serves as the error metric set for measuring prediction accuracy. Error is measured using error metric methods, including mean absolute error, root mean square error, and mean absolute percentage error, where: Let be the set of symbols associated with each error, positive for the case where the error is to be minimized, and negative otherwise; In the formula, Weights associated with each type of error; For cases where some error metrics are not differentiable, continuous processing is performed: Mean absolute error: Root mean square error: Mean absolute percentage error: T(s i A function used to measure directional changes in a time series, used to determine the consistency of the trend between predicted and actual values: s i / and i -and i-1 In the formula, s i =y i -y i-1 This represents the change in the predicted value between time i and i-1; T(s i Smooth it into a continuous, differentiable function T′(s) i ): In the formula, δ m Positive values ​​are used to avoid s i Discontinuity when δ = 0; s It is a positive value, used for numerical stability; Smoothed trend accuracy ratio p t And error ratio q t calculate: In the formula, k represents the total number of prediction times; The final formula for the smoothed error metric function is as follows: In the formula, p = p y p z +(1-p y (1-p) z ) represents the probability that the predicted trend is consistent with the actual trend, w = (2p y -1) 2 q z +(2p z -1) 2 q y +4q y q z It is an adjustment term used to correct deviations in directional matching, p y and p z These represent the growth or decline trends of the predicted and actual values, respectively. Represents the variance term; The smoothed error metric function is used as the objective function to optimize the prediction model: Objective function of the prediction combination model based on multi-criteria nonlinear programming: Constraints for load forecasting using a multi-criteria nonlinear programming-based combined forecasting model: The model is trained using historical data, and the model parameters are optimized based on error metrics and constraints.

2. The active distribution network source-load coordinated planning method considering optimal power flow as described in claim 1, characterized in that, Step 1 includes: obtaining historical load operation data samples from the power distribution network monitoring system, historical load data, weather forecasts, and distributed power generation data; cleaning and normalizing the collected load data to remove outliers and noise.

3. The active distribution network source-load coordinated planning method considering optimal power flow as described in claim 1, characterized in that, In step 3, when constructing the source-load collaborative planning model considering optimal power flow, it includes an upper-level objective function, a lower-level objective function, equality constraints, and inequality constraints, where: Upper-level objective function: Economic efficiency is the objective function. my C=C inv +C re In the formula, C inv For distributed power investment costs, C re Annual operating reliability cost; Lower-level objective function: Minimizing line loss as the objective function. In the formula, This represents the loss of line l.

4. The source-load coordinated planning method for an active distribution network considering optimal power flow as described in claim 3, characterized in that, The formula for calculating the investment cost of distributed power sources is as follows: In the formula, m WT m PV These represent the total number of candidate nodes for wind turbines and solar power, respectively. These are the unit capacity investment costs for wind turbines and solar power, respectively. The rated installed capacity of the fan at node b. The rated installed capacity of the photovoltaic system at node a; CRF is the investment recovery factor; The formula for calculating annual operational reliability cost is: In the formula, E0 is the basic electricity price of the distribution network, and T k，t Let ΔA be the actual power outage time at load point k. k，t ΔB k，t ΔEENS k，t These represent the changes in the average power outage time, average power outage frequency, and expected power shortage value for user k at load point, respectively.

5. The active distribution network source-load coordinated planning method considering optimal power flow as described in claim 3, characterized in that, Equality constraints: In the formula: P Gi Q Gi These are the active and reactive power outputs of node i, P. Di Q Di These are the active and reactive power loads at node i, V. i V j Let θ represent the voltage magnitudes at nodes i and j, where j represents all nodes directly connected to node i. ij G represents the phase angle difference between node i and node j. ij B ij These are the conductance and susceptance of the branch circuit, respectively.

6. The source-load coordinated planning method for an active distribution network considering optimal power flow as described in claim 3, characterized in that, Inequality constraints include node voltage constraints, branch current constraints, line power flow constraints, installed capacity constraints, and distributed generation penetration constraints, among which: Node voltage constraints: The operating voltage of each node is limited to a tolerance of ±5% of the rated voltage. 0.95pu≤U i ≤1.05pu Where: the operating voltage U of each node i It is limited to a tolerance of ±5% of the rated voltage; Branch current constraints: In the formula: This represents the maximum allowable current for line l; Power flow constraints on the line: In the formula, P ij To save the active power between node i and node j, Q ij To save reactive power between node i and node j, These represent the maximum and minimum apparent power values ​​between power saving i and node j, respectively. Installation capacity constraints: In the formula, P DG，i Q DG，i These represent the active and reactive power of distributed power source access node i, respectively. Let be the maximum active power that node i is allowed to connect to the distributed power source. The maximum reactive power that node i is allowed to connect to a distributed power source; Constraints on distributed power penetration: In the formula, P DG，s For the power output of the distributed power source s, P i P represents the total load requirement of node i. loss This represents the total active power loss of the distribution network.

7. The active distribution network source-load coordinated planning method considering optimal power flow as described in claim 1, characterized in that, Step 4 includes: Step 4.1: Encode the planets representing the location and capacity of the distributed power source according to the integer encoding rules. Under the constraints, randomly generate the initial population. The initial "sun" is the individual with the largest mass in the population. The smaller the objective function, the larger the mass. Step 4.2: Simulate celestial motion using the gravitational mechanism of the planetary optimization algorithm, update the position and mass of each planet, and determine the new "sun" position, which is the global optimum in the current population. Gravitational torque: The calculation formulas for each parameter are as follows: Planetary mass: a=f s -f max In the formula, a = 2 is a constant, and α = f s -f max The smaller the objective function value of a planet, the greater its mass. i，j f max f s Let represent the objective function values ​​for the i-th or j-th planet, the worst-case planet, and the Sun, respectively. The distance between planets i and j: In the formula, Let Z be the position of planets i and j in each dimension at the t-th iteration; Z is the dimension. Step 4.3, Global Search Phase: In the formula, b is a constant, and rand(0,1) is a random number between 0 and 1. Let be the position of the sun at the t-th time, and β be a coefficient that depends on M. Let be the gravitational value in the t-th iteration. The maximum gravitational value in the t-th iteration; A reconnaissance peak mechanism is introduced, using edge solutions as the search area for the bee colony. If a planet is not updated within a certain number of iterations, a new planet position is randomly generated. X' i，j =L d +rand(0,1)×(U d -L d ) In the formula, U d L d These are the upper and lower limits of the search space, respectively; Step 4.4, Local Search Phase: In the formula, c = 2, T is the maximum number of iterations, t is the current number of iterations, and g is a random number that follows a Gaussian distribution function with mean μ = 0.5 and standard deviation σ = 0.2.

Images