A device and method for multi-objective expansion planning of power distribution network considering differentiated demand

Abstract

The application relates to the field of power distribution network target planning processing, and provides a power distribution network multi-target expansion planning device and method considering differentiated demands, which comprises an electricity consumption behavior acquisition unit, a clustering unit, a feature extraction unit, a model establishment unit and a model constraint unit; the electricity consumption behavior acquisition unit comprises acquiring electricity consumption features of peak period users and demand response potential users as electricity consumption data samples; the clustering unit comprises reducing electricity consumption data samples in an electricity consumption scene and acquiring user category labels through clustering analysis; the feature extraction unit comprises adopting a Pearson correlation coefficient as an index for electricity consumption behavior feature extraction, so that a feature set with minimum redundancy is expected to be obtained; and the model establishment unit comprises establishing a power distribution network expansion planning model with optimal system economy and maximum clean energy utilization rate, thereby providing a better planning basis for power distribution network planning after considering demand response and demand response scheme definition and the like.

Description

Technical Field

[0001] This invention relates to the field of power distribution solution technology in distribution networks, and specifically to a multi-objective extended planning device and method for distribution networks that considers differentiated needs. Background Technology

[0002] In recent years, to address the ever-increasing energy demand and environmental pollution, the development of distributed generation (DG) and demand-side management have become important measures to promote the development of the global energy system. Demand response (DR), as a key technology in demand-side management, has achieved goals such as peak shaving, frequency regulation, and power balancing. However, its implementation faces the challenge of developing suitable demand response schemes based on the different electricity consumption characteristics of various users. The widespread deployment of smart metering terminals has enabled real-time collection of user data. The similarity and correlation of this data reflect users' electricity consumption habits and behavioral characteristics. Using this data, a data-driven electricity behavior profiling method can help define demand response schemes. Therefore, with the development of the electricity market and the increasing sophistication of smart metering terminals, it is necessary to study a multi-objective extended planning method for distribution networks that considers electricity behavior profiling to meet the needs of distribution network development.

[0003] Currently, load balancing (DR) is mainly divided into two categories: peak load balancing (PDR) and time-of-use load balancing (IDR). PDR generally encourages users to adjust their electricity consumption voluntarily through peak-valley time-of-use pricing, thereby achieving the effect of peak shaving and valley filling of user load in the system. IDR, on the other hand, involves signing compensation contracts with users to appropriately reduce a suitable proportion of the load when the system's power supply is insufficient.

[0004] Overall, research and application of demand-side response (DSR) schemes are relatively mature abroad, while domestic implementation is still in its early stages. The number of provinces implementing DSR is relatively small, marketization is weak, and current pilot programs primarily target industrial users with low user participation. Current theoretical research is also insufficient, with in-depth studies of actual electricity consumption patterns, resulting in DSR schemes lacking relevance to actual user energy usage and failing to provide effective incentives or regulation for electricity users. Further in-depth research is needed on how to develop DSR implementation schemes that align with my country's national conditions. Currently, many scholars are researching the application of demand-side response (DR) in distribution network expansion planning. Reference 1 [Gao Hongjun, Liu Junyong. Active distribution network collaborative planning considering different types of DG and load modeling [J]. Proceedings of the CSEE, 2016, 36(18): 4911-4922.] considers incentive-based dispatchable DR in the distribution network extension planning, and performs extension planning on the distribution network topology, DG and the capacity of loads participating in DR. Reference 2 [Lu Jun, Zhu Yanping, Peng Wenhao, Sun Yi. Smart electricity user behavior analysis feature optimization strategy [J]. Automation of Electric Power Systems, 2017, 41(05): 58-63+83.] achieves better user electricity profile classification with less computation through quantitative analysis of the effectiveness and correlation evaluation of electricity behavior features.

[0005] However, the aforementioned research literature on user electricity behavior profiling does not mention methods for analyzing and selecting electricity behavior characteristics to define user demand response solutions; at the same time, the computing speed also urgently needs to be improved in the face of rapidly increasing electricity data volume. Summary of the Invention

[0006] This invention proposes an optimal strategy for electricity consumption behavior characteristics defined by demand response schemes, and incorporates this strategy into the operational phase model establishment process of distribution network expansion planning, thereby realizing optimized expansion planning of the distribution network.

[0007] To achieve the above objectives, the technical solution adopted by this invention is: a multi-objective extended planning device for distribution networks that considers differentiated needs, comprising an electricity consumption behavior acquisition unit, a clustering unit, a feature extraction unit, a model building unit, and a model constraint unit. The electricity consumption behavior acquisition unit acquires the electricity consumption characteristics of users during peak hours and users with demand response potential as electricity consumption data samples. The clustering unit reduces the electricity consumption data samples by electricity consumption scenarios and obtains user category labels through cluster analysis. The feature extraction unit uses the Pearson correlation coefficient as an indicator for extracting electricity consumption behavior features, aiming to obtain a feature set with minimal redundancy. The model building unit establishes a distribution network extended planning model that maximizes system economics and clean energy utilization. The model constraint unit uses a normalization method and a plane constraint method to solve the constraints of the distribution network extended planning model.

[0008] Preferably, all users with demand response potential are considered as potential users eligible to participate in PDR, while only users with obvious peak-hour electricity consumption characteristics are considered as potential users eligible to participate in IDR, and users with single-peak evening electricity consumption characteristics are considered as users without demand response potential.

[0009] Preferably, the clustering analysis in the clustering unit includes performing clustering analysis on the electricity consumption data samples using an improved k-means clustering algorithm based on the clustering effectiveness index controlling the k value, in order to obtain the category label of each user, and defining the Euclidean distance from the feature vector to the cluster center as the clustering effectiveness index of user load.

[0010] On the other hand, a multi-objective extended planning method for distribution networks that considers differentiated needs uses the squared error function as the objective function for clustering units. In the formula, For the number of user classes, Total number of users For the first j Load feature vector of individual users To the i Class Center Euclidean distance, For 0-1 variables, if the first... j The user belongs to the first i The class is denoted as 1. The following constraint must also be satisfied: .

[0011] Preferably, the optimal feature set extraction includes using the electricity consumption characteristics of the user's electricity consumption curve to characterize the user's electricity consumption behavior, and selecting appropriate electricity consumption characteristics to define the user demand response scheme. The electricity consumption characteristics include daily load rate, daily peak-valley difference rate, peak power consumption rate, percentage of electricity consumption during normal periods, valley power coefficient, peak load rate, and valley load rate.

[0012] Preferably, the Pearson correlation coefficient is used as the indicator for extracting electricity consumption behavior features, with the aim of obtaining a feature set with minimal redundancy. The Pearson correlation coefficient is... In the formula, U For the optimal feature set, , These are the first in the optimal feature set. i The, the j One characteristic, For covariance, Let Variance be the variance.

[0013] Preferably, the objective function in the distribution network expansion planning model includes minimizing the overall annual investment and operation cost of the system and maximizing the DG utilization rate of the system under all scenarios within the planning year.

[0014] Preferably, the two objective functions also include multiple constraints, specifically equipment investment constraints, network radial constraints, connectivity constraints, power balance constraints, component operating characteristic constraints, PDR constraints, and IDR constraints.

[0015] Preferably, based on the above multiple constraints and objective function, the normalized planar constraint method is used for solving the problem. The solution includes... Step 91: Construct the minimization objective function based on all constraints. The single-objective optimization problem is obtained. minimum value The corresponding system DG utilization rate at this time is Let the optimal solution be... ; Step 92: Construct the minimization objective function based on all constraints. The single-objective optimization problem is obtained. minimum value The corresponding annual comprehensive cost of the system at this time is Let the optimal solution be... ; Step 93: Perform a standardization process on the objective function using the following formula: by The x-axis is... Using the vertical axis as the ordinate, construct the normalized Pareto front, utopia line, and solution space; Step 94: Click and points Obtain the Utopia line, when the Utopia line is divided into m Segments, each segment is [length missing] , for the j Dividing points Construct vectors normal vector ,vector Intersects with the Pareto front at point The optimal solution is obtained by calculating using the following formula. The formula is ; Step 95: Move the dividing point along the Utopia line Repeat step 4 to obtain a series of points uniformly distributed on the Pareto front, excluding the two endpoints, and then the Pareto solution set can be plotted.

[0016] Preferably, a multi-objective extended planning method for distribution networks that takes into account differentiated needs also includes the application of the distribution network extended planning model in multiple distribution networks and IEEE standard systems.

[0017] In summary, the beneficial effects of the present invention are as follows: 1. Good applicability: Prior to this, relevant research has been carried out at home and abroad on distribution network expansion planning, but there is almost no research on distribution network expansion planning that considers differentiated demand response schemes. This invention combines relevant research on the definition of differentiated demand response schemes and the problems encountered in distribution network expansion planning. It mainly focuses on the investment planning of distribution network lines and DG, providing a better planning basis for distribution network planning that considers demand response and the definition of demand response schemes. 2. The optimization algorithm is objective and clear: This invention develops a normalized normal constraint (NNC) optimization method to obtain the Pareto solution set, which can objectively reveal the relationship between the two objectives of DG utilization and economic cost, and the optimization algorithm is simple and clear; 3. Strong robustness: This invention can be applied in multiple power distribution networks and IEEE standard systems, and can provide reasonable planning schemes under various power supply-grid configurations. This demonstrates that this invention has wide applicability and robustness. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the demand response scheme definition process in an embodiment of the present invention; Figure 2 This is a schematic diagram of the solution process in an embodiment of the present invention; Figure 3 This is a schematic diagram of the normalized Pareto front, utopia line, and solution space in an embodiment of the present invention. Detailed Implementation

[0019] The following description, in conjunction with the appendix of the present invention, Figure 1 The technical solutions in the embodiments of the present invention are clearly and completely described herein. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Example: Currently, demand response (DR) is mainly divided into two categories: peak-valley demand response (PDR) and independent demand response (IDR). PDR generally encourages users to voluntarily adjust their electricity consumption through peak-valley time-of-use pricing, thereby achieving peak shaving and valley filling of the system's load. IDR, on the other hand, involves signing compensation contracts with users to appropriately reduce a suitable proportion of the load when the system's power supply is insufficient. To encourage user participation in DR and improve the overall operational efficiency of the distribution network, this embodiment considers all users with demand response potential as potential PDR participants, while only users with obvious peak-hour electricity consumption characteristics are considered potential IDR participants. Based on the above settings, Figure 1 The process for defining demand response solutions is given. Figure 1 Among them, users with a clear single-peak characteristic of nighttime electricity consumption are considered to be users without demand response potential.

[0021] Improved k-means clustering analysis: The first step in profiling electricity consumption behavior is to perform clustering analysis on electricity consumption data samples to reduce the electricity consumption scenarios and obtain category labels for each user. The k-means algorithm can perform clustering analysis on historical data of actual systems, avoiding model errors when fitting probability distributions to random variables. However, the choice of clustering value k has a significant impact on the clustering effect. Therefore, this invention uses an improved k-means clustering algorithm based on a clustering effectiveness index to control the value of k to perform clustering analysis on electricity consumption data samples to obtain category labels for each user. The Euclidean distance from the feature vector to the cluster center is defined as the clustering effectiveness index for user load, and its objective function is the squared criterion error function: (1) In the formula, For the number of user classes, Total number of users For the first j Load feature vector of individual users x j To the i Class Center c i Euclidean distance, For 0-1 variables, if the first... j The user belongs to the first i The class is denoted as 1. The following constraint must also be satisfied: (2) Optimal Feature Set Extraction: This invention uses the electricity consumption characteristics of user electricity consumption curves to characterize user electricity consumption behavior, and selects appropriate electricity consumption characteristics to define user demand response schemes. Commonly used electricity consumption characteristics are shown in Table 1.

[0022] Table 1 Common Electricity Consumption Characteristics This invention uses the Pearson correlation coefficient as an indicator for extracting electricity consumption behavior features, aiming to obtain a feature set with minimal redundancy. The Pearson correlation coefficient is calculated according to equation (3): (3) In the formula, U For the optimal feature set, , These are the first in the optimal feature set. i The, the j One characteristic, For covariance, Let Variance be the variance.

[0023] The minimum redundancy index is defined as: (4) In the formula, This represents the number of features contained in the optimal feature set.

[0024] Further solve for the characteristic set that satisfies equation (4) U This is the optimal feature set. This invention uses a traversal method to find the optimal feature set. Let... As a set membership indicator, Characteristic features Exists in the optimal feature set U middle, Characteristic features Not present in the optimal feature set U In the middle, equation (4) can be transformed into: (5) By traversal For all possible values, find the metric that minimizes redundancy. The largest Using vectors, we can find the optimal feature set.

[0025] User demand response scheme definition: Improved k-means clustering was used to obtain category labels for user electricity consumption behavior. The optimal feature set obtained above represents the user's electricity consumption behavior labels. This invention defines the demand response scheme for each user category by scoring each label for each user category. The maximum score is set to 100 points. i Class user corresponding to the first j The scoring formula for each tag is: (6) In the formula, For the first i Class of users j The average value of each label, For the first j The maximum value of each label. For the first j The minimum value of the nth label. When a user group (corresponding to the IDR label) scores higher than 80% of the users, that user group is considered a potential user who can participate in IDR and PDR.

[0026] Distribution network expansion planning model considering differentiated demand response schemes: The specific distribution network expansion planning model considers the optimal system economy and the maximum utilization rate of clean energy. This invention establishes a distribution network expansion planning model around this optimization objective.

[0027] Objective function 1: Minimize the system's annual comprehensive cost Considering the overall economic efficiency of the system, the objective function is to minimize the overall annual investment and operating cost of the system, which is determined by the investment costs of the newly built DG and distribution lines. Annual operating costs Composition. The relevant objective function is shown below. (7) (8) (9) (10) (11) (12) (13) (14) (15) In the formula, The annual investment cost for new DG and distribution lines is calculated. The annual operating cost of the system, The total investment cost for the new DG and distribution lines, This is the capital recovery coefficient. The annual electricity purchase cost of the distribution network, The annual incentive compensation cost for users participating in IDR, To account for the scheduling costs of participating in PDR, Annual network loss cost, Annual maintenance cost, r The discount rate is... For the first i The lifespan of each device. , , This refers to the set of candidate photovoltaic nodes, the set of candidate wind turbine nodes to be built, and the set of candidate power lines to be built. and The investment costs per unit for photovoltaic and wind turbines are respectively. The construction cost per kilometer of the line, As alternative routes Length, , , These are 0-1 variables representing investment decisions for photovoltaics, wind turbines, and power lines, respectively. D The number of days in a year. For the scene s The probability of occurrence T The number of runtime segments considered for the scenario. The interval between each time period. for t Electricity purchase price during different time periods for t The amount of electricity purchased from the main grid during a given time period. and They are respectively t The set of user nodes participating in IDR and PDR during the specified time period. for t Time period nodes a The unit compensation cost of load reduction, for t Time period nodes a The amount of load reduction, for t Initial electricity price for the period for t Time period nodes b The initial power of the load, for t Time period nodes b Changes in electricity prices after the implementation of PDR for t Time period nodes b After implementing PDR, the load's response power is based on the electricity price signal. The unit network loss coefficient, For the active power loss of the distribution network, This represents the equipment maintenance cost coefficient.

[0028] Objective function 2: Maximize system DG utilization This invention uses maximizing the DG utilization rate of the system under all scenarios within the planned year as another objective function for the expansion planning of the distribution network.

[0029] (16) In the formula, and These represent the power generation capacity of wind turbines and photovoltaic systems, respectively.

[0030] Constraints: 1) Equipment investment constraints Assuming the maximum number of photovoltaic and wind turbines and the number of newly built lines in the planned distribution network are predetermined, the equipment investment constraints for the distribution network expansion plan can be expressed as: (17) In the formula, , These refer to the maximum number of newly built wind turbines and photovoltaic power plants, respectively. This represents the projected number of new lines to be built.

[0031] 2) Network radial constraints and connectivity constraints In the expansion planning of power distribution lines, the formation of ring structures should be avoided, and the principle of "closed-loop design, open-loop operation" should be met. Under the premise of ensuring connectivity constraints, the simplest constraint is that the final planned number of lines is one less than the number of nodes. Therefore, the radial constraint of the network in this invention can be expressed as: (18) In the formula, n and m These represent the total number of nodes and the total number of lines in the planned power distribution network.

[0032] At the same time, in order to ensure the connectivity of each node in the planned power grid, it is necessary to verify the connectivity of the topology of the planned distribution network.

[0033] 3) Power balance constraints The planned power distribution network should meet power balance constraints. Equations (19)-(21) represent the node power balance constraints, and equations (22) and (23) are used to determine the power balance constraints of the nodes. M The law's line flow constraint is achieved by introducing a sufficiently large positive real number. M Make the line l When not in operation, inequality constraints (22) and (23) are invalid, while (24) and (25) are upper and lower limits of line power constraints.

[0034] (twenty one) (twenty two) (twenty three) (twenty four) (25) In the formula, Forj The collection of routes at the beginning, For j For the terminal line set, For the line ij The square of the current value, For nodes i The square of the voltage value, and The lines are respectively ij Active and reactive power, For the line ij The resistance value, For the line ij The reactance value, , These represent the reactive power purchased from and sold to the main grid, respectively. , , These are the reactive power output of wind turbines, reactive power output of photovoltaic systems, and reactive power consumed by the load. For a sufficiently large positive real number, and The lines are respectively ij The maximum active and reactive power transmitted.

[0035] 4) Component operating characteristic constraints a. Operational constraints of photovoltaic power generation (26) In the formula, For the first i The upper limit of the power generation capacity of a single photovoltaic unit.

[0036] b. Wind turbine operating constraints (27) In the formula, No. i The upper limit of the power generation capacity of a single wind turbine unit.

[0037] 5) PDR constraint PDR encourages users to voluntarily adjust their electricity consumption by setting reasonable day-ahead electricity prices, so its response is closely related to electricity prices. Equation (28) shows the relationship between the rate of change of electricity prices and the rate of change of load.

[0038] (28) In the formula, for t Load change rate over time period for t Electricity price change rate over time period EThis is the demand elasticity matrix for PDR, with the main diagonal elements being the self-elasticity coefficients and the secondary diagonal elements being the reciprocal elasticity coefficients.

[0039] Therefore, the equality constraint between load power before and after implementing PDR can be expressed as: (29) Furthermore, to prevent excessive load response during PDR from affecting users' normal electricity consumption, the total load should remain unchanged before and after PDR implementation, i.e.: (30) At the same time, the rights and interests of users participating in PDR also need to be considered. Therefore, this invention introduces constraints on electricity usage satisfaction and electricity cost satisfaction to protect user rights.

[0040] (31) In the formula, The minimum satisfaction level with users' electricity usage methods. The minimum value for user satisfaction with electricity costs.

[0041] 6) IDR constraint This invention considers IDR primarily in the form of load reduction, requiring consideration of the response constraints of participating users based on the characteristics of load reduction responses. To avoid additional adjustment costs during IDR implementation, the amount of load reduction response submitted by users should meet certain constraints, namely: (32) In the formula, This is the percentage of load that can be reduced during IDR implementation; its value depends on the type and nature of the load.

[0042] Meanwhile, the total load reduction during the entire scheduling cycle should not exceed the pre-defined maximum load reduction capacity.

[0043] (33) In the formula, Reduce capacity to the pre-defined maximum load.

[0044] Model transformation: The extended planning model of the distribution network constructed in this invention is a mixed integer nonlinear model. The node power balance constraints and objective function in the model are nonconvex and nonlinear. The nonconvex terms in the node power balance constraints and objective function are processed by second-order cone relaxation and piecewise linear processing, respectively.

[0045] In the extended planning model of the distribution network constructed in this invention, the non-convex terms in the node power balance constraint are mainly the product of the second-order terms of current and voltage, and the square terms of active and reactive power. Equation (21) is convexly relaxed and then transformed into the following standard second-order cone form.

[0046] (34) In the formula, Given the Euclidean norm, the objective function of this invention is: The increasing function satisfies the applicable conditions of the second-order cone constraint, and the second-order cone relaxation must be accurate.

[0047] In the objective function 1 of this invention, the nonlinear term in equation (13) mainly comes from the product of the change in electricity price and the response power after the implementation of PDR. This invention uses the triangle method to handle this nonlinear term, which is essentially a piecewise linearization method for bivariate variables. Equation (13) indicates that the dispatch cost of participating in PDR is related to two factors: the change in electricity price and the response power, and can be expressed as: (35) Considering On the shaft common X + 1 breakpoint, while On the shaft common Y + 1 breakpoint. Through segmentation, the surface formed by equation (35) is divided into... X × Y A small rectangle. Further consideration is given by intervals. and ( x = 1, 2, …, X ; y = 1, 2, …, Y A rectangle formed by ) is introduced, with continuous variables. and 0-1 variables , , respectively, correspond to the upper and lower triangles of the rectangle. Then, the PDR scheduling cost corresponding to equation (35) can be approximated using the triangle method with the following constraints: (36) (37) In the formula, and The boundary value is .

[0048] Through the above steps, the extended planning model of the power distribution network established in this invention has been transformed into a mixed integer linear programming problem, which can be solved quickly using mature commercial software or algorithm packages (CPLEX, GUROBI).

[0049] Model Solution Process: To efficiently solve the above model, this invention employs the normalized normal constraint (NNC) method to solve the multi-objective optimization problem. The basic process is as follows: Figure 2 As shown. Compared to other multi-objective optimization methods, the NNC method can transform the multi-objective optimization problem into solving the optimal Pareto front problem corresponding to each equidistant split point. Thus, the complete Pareto front can be described by finding fewer Pareto front points, which can significantly reduce the amount of computation.

[0050] The main steps of the above algorithm are explained below: Step 1: Construct the objective function to minimize using equations (17) to (37) as constraints. The single-objective optimization problem is obtained. minimum value The corresponding system DG utilization rate at this time is Let the optimal solution be... .

[0051] Step 2: Using equations (17) to (37) as constraints, construct the objective function to minimize it. The single-objective optimization problem is obtained. minimum value The corresponding annual comprehensive cost of the system at this time is Let the optimal solution be... .

[0052] Step 3: Normalize the objective function using equation (38): (38) by The x-axis is... Using the ordinate, construct the normalized Pareto front, utopia line, and solution space. For example... Figure 3 As shown. Points on the normalized plane. and points Then the points obtained in steps 1 and 2 respectively and points The line segment connecting these two points That would be the Utopia Line.

[0053] Step 4: Assume the Utopia Line is divided into m Segments, and the length of each segment is , for thej Dividing points Construct vectors normal vector ,vector Intersects with the Pareto front at point Define the origin pointing to The vector is The origin points to The vector is By solving the single-objective optimization problem as shown in equation (39), the normal vector can be obtained. Intersection with the Pareto front .

[0054] (39) Step 5: Obtain the optimal solution using equation (39) And continuously move the dividing point along the Utopia line. Repeat step 4 to obtain a series of points uniformly distributed on the Pareto front, excluding the two endpoints, and then the Pareto solution set can be plotted.

[0055] It is worth noting that a demand response scheme for differentiated user electricity consumption characteristics is defined through an electricity consumption behavior feature optimization strategy based on improved k-means clustering analysis and optimal feature set extraction. Then, a multi-objective extended planning model for the distribution network considering differentiated demand response schemes is constructed, adding annual comprehensive cost and DG utilization rate as optimization objectives to the decision objectives, while also considering IDR and PDR constraints in the model constraints. To address the non-convex and nonlinear characteristics of this model, a second-order cone relaxation method and a triangle-based linearization method are used to transform it into a mixed-integer linear model.

[0056] In summary, this application primarily protects the analysis and selection method of electricity consumption behavior characteristics for defining user demand response schemes, and the distribution network expansion planning method considering differentiated demand response schemes.

[0057] In the description of this invention, it should be understood that the terms "counterclockwise," "clockwise," "longitudinal," "lateral," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

Claims

1. A multi-objective extended planning device for distribution networks that considers differentiated needs, characterized in that, The device includes a computer server, wherein the computer media within the computer server includes a power consumption behavior acquisition unit, a clustering unit, a feature extraction unit, a model building unit, and a model constraint unit. The electricity consumption behavior acquisition unit includes acquiring the electricity consumption characteristics of users during peak hours and users with demand response potential as electricity consumption data samples. The clustering unit includes reducing the electricity consumption scenarios of the electricity consumption data samples and obtaining user category labels through cluster analysis. The feature extraction unit includes using the Pearson correlation coefficient as an indicator for extracting electricity consumption behavior features, aiming to obtain a feature set with minimal redundancy. The model building unit includes establishing a distribution network expansion planning model with optimal system economy and maximum clean energy utilization. The model constraint unit includes using the normalization method and the plane constraint method to solve the constraints of the distribution network expansion planning model. The clustering analysis in the clustering unit includes an improved k-means clustering algorithm based on the clustering effectiveness index to control the k value, which is used to perform clustering analysis on the electricity consumption data sample to obtain the category label of each user. The Euclidean distance from the feature vector to the cluster center is defined as the clustering effectiveness index of user load. An improved k-means clustering algorithm, based on a clustering effectiveness index controlling the k-value, is used to perform clustering analysis on electricity consumption data samples to obtain category labels for each user. The Euclidean distance from the feature vector to the cluster center is defined as the clustering effectiveness index for user load. The objective function of the clustering unit is the squared error function, specifically... In the formula, For the number of user classes, Total number of users For the first j Load feature vector of individual users x j To the i Class Center c i Euclidean distance, For 0-1 variables, if the first... j The user belongs to the first i The class is denoted as 1. The following constraint must also be satisfied: The objective functions in the power distribution network expansion planning model include minimizing the comprehensive annual investment and operation cost of the system and maximizing the DG utilization rate of the system under all scenarios within the planning year. The two objective functions also include several constraints, specifically equipment investment constraints, network radial constraints, connectivity constraints, power balance constraints, component operating characteristic constraints, PDR constraints, and IDR constraints. Based on the aforementioned constraints and objective function, the normalization method and plane constraint method are used to solve the problem. The solution includes... Step 91: Construct the minimization objective function based on all constraints. f The single-objective optimization problem of 1 yields the following results: f Minimum value of 1 f 1,min The corresponding system DG utilization rate at this time is f 2,max Let the optimal solution be... ; Step 92: Construct the minimization objective function based on all constraints. f The single-objective optimization problem of 2 is obtained. f Minimum value of 2 The corresponding annual comprehensive cost of the system at this time is f 1,max Let the optimal solution be... ; Step 93: Perform a standardization process on the objective function using the following formula: ； by The x-axis is... Using the vertical axis as the ordinate, construct the normalized Pareto front, utopia line, and solution space; Step 94: Click and points Obtain the Utopia line, when the Utopia line is divided into m Segments, each segment is [length missing] , for the j Dividing points Construct vectors normal vector ,vector Intersects with the Pareto front at point The optimal solution is obtained by calculating using the following formula. The formula is ； Step 95: Move the dividing point along the Utopia line Repeat step 94 to obtain a series of points uniformly distributed on the Pareto front, excluding the two endpoints, and then the Pareto solution set can be plotted.

2. The multi-objective extended planning device for distribution networks considering differentiated needs according to claim 1, characterized in that, Users with demand response potential are considered potential users for PDR, while users with peak-hour electricity consumption characteristics are potential users for IDR, and users with single-peak evening electricity consumption characteristics are not users with demand response potential.

3. The multi-objective extended planning device for distribution networks considering differentiated needs according to claim 1, characterized in that, The optimal feature set extraction includes using the electricity consumption characteristics of user electricity consumption curves to characterize user electricity consumption behavior, selecting appropriate electricity consumption characteristics to define user demand response schemes, and electricity consumption characteristics including daily load factor, daily peak-valley difference rate, peak power consumption rate, percentage of electricity consumption during normal periods, valley power factor, peak load factor, and valley load factor.

4. The multi-objective extended planning device for distribution networks considering differentiated needs according to claim 1, characterized in that, Pearson correlation coefficient is used as the indicator for extracting electricity consumption behavior features, with the aim of obtaining a feature set with minimal redundancy. The Pearson correlation coefficient is... In the formula, U For the optimal feature set, , These are the first in the optimal feature set. i The, the j One characteristic, For covariance, Let Variance be the variance.

5. The multi-objective extended planning device for distribution networks that considers differentiated needs as described in claim 4 further includes the use of the extended planning model for distribution networks in multiple distribution networks and IEEE standard systems.

Images