Distribution network flexibility attribution analysis method based on distributed energy access

Abstract

The invention relates to a power distribution network flexibility attribution analysis method based on distributed energy access. The method comprises the following steps: constructing a scene sample of correlation between distributed energy by a Copula modeling and LHSD sampling method; extracting typical scene samples by adopting a clustering method based on a hierarchical idea; performing evaluation and attribution analysis on typical scene samples based on the feasibility of the power distribution network economic dispatching optimization model to provide effective information for selection of a power distribution network flexibility improvement method; the K-means algorithm based on the layering idea can be used for extracting typical scene samples, so that a more representative distributed energy output scene can be obtained; under an economic dispatching model of the power distribution network, the flexibility attribution analysis can provide specific causes of insufficient flexibility and potential weak links for typical scenes of distributed energy access, which is more beneficial to making decisions for control and optimization of the power distribution network; the method has the advantages that the scene samples are reduced by adopting the clustering method, the scene samples are evaluated and subjected to attribution analysis, and the flexibility of the power distribution network is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy systems and electric power engineering, and specifically relates to a distribution network flexibility attribution analysis method based on distributed energy access. Background Art

[0002] Distributed energy access refers to the access of renewable energy or other distributed energy sources to the traditional distribution network system. This access method has many advantages, such as reducing dependence on traditional power grids, improving the reliability and flexibility of power grids, etc. The continuous access of distributed energy is the development trend of distribution networks and one of the important measures taken by the power industry to solve environmental problems and energy crises; however, the penetration of distributed energy and its own randomness have an unignorable impact on the operating status of distribution networks, which makes the operators of distribution networks pay more attention to the operating safety of distribution networks, and pay more attention to the flexibility of distribution networks considering distributed energy access. In view of the intermittent nature, uncertainty and correlation between distributed energy, considering multiple scenarios has become an important link in the process of studying distribution network problems. Among them, scenario generation method and scenario analysis method are the main research technical means. Scenario generation method includes Monte Carlo sampling method, prediction scenario generation method and optimization scenario generation method; and scenario analysis method can be divided into simulation scenario timing analysis method, analysis method based on scenario output characteristics, and scenario analysis method based on clustering algorithm; At present, the flexibility of distribution network has become one of the important factors to be considered in the research on new energy power planning, energy storage optimization configuration, scheduling, etc. of distribution network, as well as in the formulation of new energy consumption strategy. By adopting the method of coordinating flexibility resources, it is conducive to proposing methods and strategies that are more suitable for the development of modern distribution network; Therefore, the research on the flexibility of distribution network itself focuses on two parts: distribution network flexibility evaluation and improvement methods; In the definition and quantification of distribution network flexibility, due to the network structure, multi-load and complex factors affecting its operating status, the definition of distribution network flexibility in domestic and foreign research has not yet been unified, but the distribution network flexibility evaluation indicators generally include four basic elements: time, resources, uncertainty and cost; Therefore, it is very necessary to provide a distribution network flexibility attribution analysis method based on distributed energy access, which adopts clustering method to reduce scenario samples, evaluate and attribution analyze scenario samples, and improve distribution network flexibility. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and to provide a distribution network flexibility attribution analysis method based on distributed energy access, which adopts a clustering method to reduce scene samples, evaluate and analyze scene samples, and achieve improved distribution network flexibility.

[0004] The object of the present invention is achieved by: a distribution network flexibility attribution analysis method based on distributed energy access, the method comprising the following steps:

[0005] Step 1: A power scenario generation tool consisting of Copula modeling and LHSD sampling method, which can be used to effectively model the uncertainty of distributed energy;

[0006] Step 2: Using a clustering method based on hierarchical thinking, the output scenario samples generated in step 1 are effectively reduced to obtain more representative output scenarios;

[0007] Step 3: After obtaining typical scenarios through the hierarchical K-means clustering algorithm in step 2, a scheduling optimization model that takes into account flexibility resources and wind turbine access is constructed;

[0008] Step 4: Conduct flexibility attribution analysis based on the distribution network economic dispatch optimization model constructed in Step 3 to understand the specific reasons for insufficient flexibility and potential weak links, and provide information for users to choose effective regulation strategies to improve flexibility.

[0009] The output scenario generation tool composed of Copula modeling and LHSD sampling method in step 1 can effectively model the uncertainty of distributed energy by using the output scenario generation tool, specifically: Copula modeling is a statistical modeling method used to describe the dependency relationship between random variables; in Copula modeling, marginal distribution and dependency structure are modeled separately to more accurately describe the dependency relationship between variables; LHSD sampling method is a statistical method for designing parameter space sampling, which is an extension of LHS. LHSD is usually used with Copula to generate samples consistent with the joint distribution; using them to form an output scenario generation tool to perform scenario generation, specifically including the following steps:

[0010] Step 1.1: Initialization: Prepare historical measurements and data of different renewable energy sources and input them into the tool, where the marginal distribution functions are usually presented as histograms;

[0011] Step 1.2: Dependency modeling: Assuming the interdependence structure, a multivariate joint distribution is constructed to represent the interdependence between renewable energy sources using linkage relations; this contains information about the degree of dependence and the dependency structure;

[0012] Step 1.3: Ranking and screening: Calculate the index of goodness of fit using Copula and select the best correlation model in the sampling stage;

[0013] Step 1.4: Scenario generation: Based on the multivariate joint distribution and dependency model obtained in the previous stage, LHSD is applied to generate scenarios;

[0014] Step 1.5: Output Analysis: Analyze the generated scenario.

[0015] The step 2 uses a clustering method based on hierarchical thinking to effectively reduce the output scenario samples generated in step 1 to obtain more representative output scenarios. Specifically, the clustering algorithm based on hierarchical thinking is a variation of the K-means clustering method. This method evaluates similarity based on sample distance and takes the sum of the distances between each sample and the center point of its category as the optimization target. This method can further improve the efficiency and accuracy of the K-means method and apply it to the reduction of distributed energy output scenarios. The implementation of the K-means algorithm based on hierarchical thinking specifically includes two stages: building a hierarchical data set and clustering layer by layer.

[0016] The first stage of the K-means algorithm based on the hierarchical idea in step 2 is specifically: the hierarchical data set is built "from top to bottom", and its purpose is to simplify the original data set layer by layer; the specific content of this stage is divided into two steps. First, in order to quickly reduce the size of the upper data set, half of the total number of individuals in the upper data set can be used as the number of groups in the lower layer, and then the K-means algorithm is used to perform clustering with the set number of groups; secondly, in order to ensure that the similarity between the simplified data set and the upper data set meets the requirements, the groups obtained by the K-means algorithm are screened for "out-of-group individuals", and the screening basis is to evaluate the similarity between the individuals in each group and the center, that is: Among them, p is a member in the group; c is the central individual of the group; ε is the similarity between the member and the central individual of the group; the user can set the specific similarity limit according to the type and scale of the data set. When the corresponding similarity of the member does not meet the requirements, it is judged as an "outlier individual". Then, the center of the group obtained by clustering by the K-means algorithm and the "outlier individual" obtained by screening together constitute the simplified data set of the lower layer. As the number of layers of the hierarchical data set increases, the similarity between the simplified data set and the original data set gradually decreases, and then the stopping condition of the hierarchical data set is constructed.

[0017] The second stage of the K-means algorithm based on the hierarchical idea in step 2 is specifically as follows: the second stage uses the K-means algorithm to perform clustering layer by layer "from bottom to top"; first, from the simplified data set of the lowest layer, the initial center point is selected according to the expected number of thing groups, which is more in line with the ideal clustering requirements than the quality of the randomly selected initial center point of the traditional K-means algorithm; secondly, the new center point obtained by clustering the simplified data set of the lower layer through the K-means algorithm will be used as the initial center point of the clustering of the data set of the upper layer.

[0018] After obtaining a typical scenario through the hierarchical K-means clustering algorithm in step 2 in step 3, a scheduling optimization model taking into account flexibility resources and wind turbine access is constructed. Specifically, the model is based on the maximum acceptance scheduling with the minimum wind abandonment, that is, under the premise of ensuring the safety and stability of the system, the wind abandonment of wind power is minimized through optimized scheduling; at the same time, the economic efficiency of utilizing flexibility resources is considered, in which the main transformer and interruptible load of the distribution network are regarded as available flexibility resources, and the objective function of the model includes the penalty cost function of wind abandonment, the power purchase cost, and the interruptible load compensation fee; then the model of the objective function is: Where P P,t C is the amount of electricity purchased; P is the unit electricity purchase cost; P W,t is the abandoned air volume; C W P is the penalty cost per unit wind abandonment; L,t is the load shedding capacity; C L P is the penalty cost per unit load shedding; IL,t is the interruptible load reduction; C IL is the compensation cost per unit interruptible load; T represents the time period; the constraints of the optimization model include system operation constraints and flexibility resource constraints.

[0019] The system operation constraints of the scheduling optimization model taking into account the flexibility resources and wind turbine access in step 3 include: ① The power flow balance equation constraint refers to the calculation equation of the active and reactive power injected into the node, which can be expressed in the following compact form: f(y,u)=0(3), where y is the state variable, i.e., the voltage amplitude and phase angle of the unbalanced node; u is the control vector; ② The voltage constraint refers to the upper and lower voltage limits that the operating voltage must meet, which can be expressed as: V min ≤V i ≤V max (4), where V i represents the voltage amplitude of the ith unbalanced node; V min 、V max are the maximum and minimum values ​​of the node voltage respectively; ③ Transmission limit constraint means that each line has a certain transmission capacity, and the operating transmission current must be less than the maximum line transmission limit to ensure transmission safety: |I ij (y,u)|≤I ij,max (5), where I ij I represents the current flowing through the line from node i to node j; ij,max is the maximum current amplitude that the line can carry; ④ Distributed energy output power constraint: P Gi,min ≤P i (y,u)≤P Gi,max (6), QGi,min ≤Q i (y,u)≤Q Gi,max (7), where P Gi,min , P Gi,max They represent the minimum and maximum values ​​of the distributed energy output active power of the i-th node respectively; Q Gi,min , Q Gi,max Output the minimum and maximum values ​​of reactive power respectively; P i (y,u),Q i (y,u) are respectively the active and reactive power output of the distributed energy of the ith node under current operation; ⑤ Wind farm dispatch power constraint: 0≤P Gi (y,u)≤P Gi,max (8), where P Gi,max represents the distributed energy active output of the i-th node; P Gi (y,u) represents the active power actually injected into the grid by the distributed energy connected to the i-th node; ⑥ Wind abandonment constraint, that is, the wind power generation power abandoned is lower than: 0≤ΔP w,t ≤P w,t (9), where ΔP w,t is the abandoned air volume; P w,t is the wind power generation power.

[0020] The flexibility resource constraints of the scheduling optimization model taking into account the flexibility resources and wind turbine access in step 3 include: ① Transformer margin constraint, which means that the flexibility provided by the transformer is lower than the flexibility margin of the transformer, as shown in the following formula: P Tmin ≤ΔP T ≤P Tmax (10), where ΔP T Represents the active power provided by the transformer; P Tmin Represents the minimum active power that the transformer can provide; P Tmax is the maximum active power that can be provided; ② interruptible load constraint, which means that the flexibility provided by the interruptible load must be lower than the maximum flexibility of the interruptible load, as shown in the following formula: 0≤P IL ≤P ILMax (11), where P IL Represents the active power provided by the interruptible load; P ILMax Represents the maximum active power that the interruptible load can provide.

[0021] In step 4, flexibility attribution analysis is performed on the basis of the distribution network economic dispatch optimization model constructed in step 3, so as to understand the specific reasons for insufficient flexibility and potential weak links, and provide information for users to choose effective control strategies for improving flexibility. Specifically, based on the distribution network economic dispatch optimization model, flexibility attribution analysis is performed. The core idea is to analyze the existence of feasible solutions for different scenarios under the above constraints. At the same time, the equality and inequality constraints (3) to (11) can be converted into equality constraints by using the relaxation variable technology, and the constraint set of the economic dispatch problem of considering flexibility resources can be finally expressed in the following abbreviated form: H(x) = 0 (12), and the field is constrained according to the satisfaction of the constraint set (12). Scenario screening, that is, solving the feasible solutions of constraint set (12) under each scenario, and dividing the typical output scenario into feasible scenarios and infeasible scenarios according to the constraint satisfaction of the solution results; after solving the typical output scenario by the method of fast calculation of feasible solutions based on dynamic trajectories, when the typical output scenario is infeasible under the condition of constraint set (12), further detailed analysis is conducted on the over-limit situation of the solution of the infeasible scenario under constraint set (12), and the constraints with the largest over-limit amount are classified to determine the type of cause leading to the insufficient flexibility of the distribution network system; then, for the feasible output scenario, further analysis is conducted on the feasible scenarios close to over-limit in constraint set (12), and then the categories of the constraints close to over-limit are distinguished and the cause of the potential insufficient flexibility is determined;

[0022] Therefore, typical scenarios are analyzed into three categories: feasible scenarios refer to scenarios that satisfy constraint set (12), that is, H(x) = 0; infeasible scenarios refer to scenarios that do not satisfy constraint set (12), that is, H(x) ≠ 0; critical scenarios refer to scenarios in which some constraints in the solution do not exceed the limit, but the distance from the limit value of the corresponding constraint and the ratio of the constraint range are less than the set value.

[0023] Beneficial effects of the present invention: The present invention is a distribution network flexibility attribution analysis method based on distributed energy access. In use, the present invention comprehensively considers the volatility and correlation of distributed energy, and the flexibility of the distribution network is correlated with the optimization application of the distribution network. For the distribution network with distributed energy access, a clustering method based on the hierarchical idea is used to reduce the generated distributed energy output scenarios, and a distribution network economic dispatch optimization model considering flexibility is used to screen typical scenarios. Then, from the perspective of the mathematical model of the distribution network economic dispatch optimization application, based on the feasibility of the distribution network economic dispatch optimization model, typical scenario samples are evaluated and attributed. The reasons for insufficient or lack of adjustment flexibility in the distribution network application caused by the access of distributed energy are analyzed, so as to provide effective information for the selection of methods to improve the flexibility of the distribution network; the present invention can provide specific reasons for insufficient flexibility and potential weak links for typical scenarios of distributed energy access through flexibility attribution analysis, and this information is more conducive to operators making decisions for distribution network control and optimization; the present invention has the advantages of using a clustering method to reduce scenario samples, evaluate and attribute the scenario samples, and improve the flexibility of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the present invention.

[0025] Figure 2 This is a flow chart of the scene generation method of the present invention.

[0026] Figure 3 This is a flow chart of the K-means algorithm based on the hierarchical concept of the present invention.

[0027] Figure 4 This is a flow chart for scene generation and typical scene extraction of distributed energy in the present invention.

[0028] Figure 5 This is a schematic diagram of distributed energy scenario evaluation and analysis of the present invention.

[0029] Figure 6 It is a 14-node structure diagram and a schematic diagram of the distributed energy access location of the present invention.

[0030] Figure 7 This is a schematic diagram of a typical scenario based on the K-means clustering algorithm of the present invention.

[0031] Figure 8 A schematic diagram of a typical scenario of the hierarchical K-means clustering algorithm of the present invention.

[0032] Fig. 9 This is a diagram of the maximum limit-crossing constraint analysis results under the initial sampling scenarios (1000) of the present invention.

[0033] Fig.10 This is a diagram of the maximum limit-crossing constraint analysis results in typical scenarios (100) after clustering of the present invention. DETAILED DESCRIPTION

[0034] The present invention will be further described below in conjunction with the accompanying drawings.

[0035] Example 1

[0036] like Figure 1-10 As shown, a distribution network flexibility attribution analysis method based on distributed energy access, the method comprises the following steps:

[0037] Step 1: A power scenario generation tool consisting of Copula modeling and LHSD sampling method, which can be used to effectively model the uncertainty of distributed energy;

[0038] Step 2: Using a clustering method based on hierarchical thinking, the output scenario samples generated in step 1 are effectively reduced to obtain more representative output scenarios;

[0039] Step 3: After obtaining typical scenarios through the hierarchical K-means clustering algorithm in step 2, a scheduling optimization model that takes into account flexibility resources and wind turbine access is constructed;

[0040] Step 4: Conduct flexibility attribution analysis based on the distribution network economic dispatch optimization model constructed in Step 3 to understand the specific reasons for insufficient flexibility and potential weak links, and provide information for users to choose effective regulation strategies to improve flexibility.

[0041] The present invention is a distribution network flexibility attribution analysis method based on distributed energy access. In use, the present invention comprehensively considers the volatility and correlation of distributed energy, and the flexibility of the distribution network is correlated with the distribution network optimization application. For the distribution network with distributed energy access, a clustering method based on the hierarchical idea is used to reduce the generated distributed energy output scenarios, and a distribution network economic dispatch optimization model considering flexibility is used to screen typical scenarios. Then, from the perspective of the mathematical model of the distribution network economic dispatch optimization application, based on the feasibility of the distribution network economic dispatch optimization model, typical scenario samples are evaluated and attributed. The reasons for insufficient or lack of adjustment flexibility in the distribution network application caused by the access of distributed energy are analyzed, so as to provide effective information for the selection of the distribution network flexibility improvement method; the present invention can provide specific reasons for insufficient flexibility and potential weak links for typical scenarios of distributed energy access through flexibility attribution analysis, and this information is more conducive to the operation personnel to make decisions for the control and optimization of the distribution network; the present invention has the advantages of using a clustering method to reduce scenario samples, evaluate and attribute the scenario samples, and improve the flexibility of the distribution network.

[0042] Example 2

[0043] like Figure 1-10 As shown, a distribution network flexibility attribution analysis method based on distributed energy access, the method comprises the following steps:

[0044] Step 1: A power scenario generation tool consisting of Copula modeling and LHSD sampling method, which can be used to effectively model the uncertainty of distributed energy;

[0045] Step 2: Using a clustering method based on hierarchical thinking, the output scenario samples generated in step 1 are effectively reduced to obtain more representative output scenarios;

[0046] Step 3: After obtaining typical scenarios through the hierarchical K-means clustering algorithm in step 2, a scheduling optimization model that takes into account flexibility resources and wind turbine access is constructed;

[0047] Step 4: Conduct flexibility attribution analysis based on the distribution network economic dispatch optimization model constructed in Step 3 to understand the specific reasons for insufficient flexibility and potential weak links, and provide information for users to choose effective regulation strategies to improve flexibility.

[0048] Figure 2 Flowchart of the proposed scenario generation method, Figure 2The Copula modeling proposed in is a statistical modeling method used to describe the dependencies between random variables. In Copula modeling, the marginal distribution and the dependency structure are modeled separately to more accurately describe the dependencies between variables. The LHSD sampling method is a statistical method for designing parameter space sampling. It is an extension of LHS. LHSD is usually used with Copula to generate samples consistent with the joint distribution. They are used to form a scenario generation tool to generate scenarios, which specifically includes the following steps:

[0049] Step 1.1: Initialization: Prepare historical measurements and data of different renewable energy sources and input them into the tool, where the marginal distribution functions are usually presented as histograms;

[0050] Step 1.2: Dependency modeling: Assuming the interdependence structure, a multivariate joint distribution is constructed to represent the interdependence between renewable energy sources using linkage relations; this contains information about the degree of dependence and the dependency structure;

[0051] Step 1.3: Ranking and screening: Calculate the indicators used to measure the goodness of fit using Copula (Spearman rank when studying spatial correlation, Euclidean distance between given distributions, two-sample test, etc.), and select the best correlation model in the sampling stage;

[0052] Step 1.4: Scenario generation: Based on the multivariate joint distribution and dependency model obtained in the previous stage, LHSD is applied to generate scenarios;

[0053] Step 1.5: Output Analysis: Analyze the generated scenario.

[0054] Figure 3 This is a flowchart of the K-means algorithm based on hierarchical thinking. Figure 3 The clustering algorithm based on hierarchical thinking proposed in the paper is a variation of the K-means clustering method. This method evaluates similarity based on sample distance and takes the sum of the distances between each sample and the center point of its category as the optimization target. This method can further improve the efficiency and accuracy of the K-means method and apply it to the reduction of distributed energy output scenarios. The implementation of the K-means algorithm based on hierarchical thinking specifically includes two stages: building a hierarchical data set and clustering layer by layer.

[0055] The first stage is as follows: the hierarchical data set is built "from top to bottom", and its purpose is to simplify the original data set layer by layer; the specific content of this stage is divided into two steps. First, in order to quickly reduce the size of the upper data set, half of the total number of individuals in the upper data set can be used as the number of groups in the lower layer, and then the K-means algorithm is used to cluster with the set number of groups; secondly, in order to ensure that the similarity between the simplified data set and the upper data set meets the requirements, the groups obtained by the K-means algorithm are screened for "out-of-group individuals". The screening basis is to evaluate the similarity between the individuals in each group and the center, that is: Among them, p is a member in the group; c is the central individual of the group; ε is the similarity between the member and the central individual of the group; the user can set the specific similarity limit according to the type and scale of the data set. When the corresponding similarity of the member does not meet the requirements, it is judged as an "outlier individual". Then, the center of the group obtained by clustering by the K-means algorithm and the "outlier individual" obtained by screening together constitute the simplified data set of the lower layer. As the number of layers of the hierarchical data set increases, the similarity between the simplified data set and the original data set gradually decreases, and then the stopping condition of the hierarchical data set is constructed.

[0056] The second stage is as follows: The second stage uses the K-means algorithm to perform clustering layer by layer "from bottom to top"; first, from the simplified data set of the lowest layer, the initial center point is selected according to the expected number of groupings of things. Compared with the quality of the randomly selected initial center point of the traditional K-means algorithm, it is more in line with the ideal clustering requirements; secondly, the new center point obtained by clustering the simplified data set of the lower layer through the K-means algorithm will be used as the initial center point of the clustering of the data set of the upper layer.

[0057] Figure 4 Generate scenarios for distributed energy and extract flowcharts for typical scenarios. Figure 4 is Figure 2 On the basis of the above, a hierarchical K-means clustering algorithm is introduced. Through this algorithm, the generated output scenario samples are effectively reduced to obtain more representative output scenarios. Specifically, the model is based on the maximum acceptance dispatch with the minimum wind abandonment, that is, under the premise of ensuring the safety and stability of the system, the wind power abandonment is minimized by optimizing the dispatch; at the same time, the economic efficiency of utilizing flexible resources is considered, in which the main transformer and interruptible load of the distribution network are regarded as available flexible resources. The objective function of the model includes the penalty cost function of wind abandonment, the power purchase cost, and the compensation cost of interruptible load; the model of the objective function is: Where P P,t C is the amount of electricity purchased; P is the unit electricity purchase cost; P W,t is the abandoned air volume; C W P is the penalty cost per unit wind abandonment;L,t is the load shedding capacity; C L P is the penalty cost per unit load shedding; IL,t is the interruptible load reduction; C IL is the compensation cost per unit interruptible load; T represents the time period; the constraints of the optimization model include system operation constraints and flexibility resource constraints.

[0058] The system operation constraints include: ① The power flow balance equation constraint refers to the calculation equation of the active and reactive power injected into the node, which can be expressed in the following compact form: f(y,u)=0(3), where y is the state variable, i.e., the voltage amplitude and phase angle of the unbalanced node; u is the control vector; ② The voltage constraint refers to the upper and lower voltage limits that the operating voltage must meet, which can be expressed as: V min ≤V i ≤V max (4), where V i represents the voltage amplitude of the ith unbalanced node; V min 、V max are the maximum and minimum values ​​of the node voltage respectively; ③ Transmission limit constraint means that each line has a certain transmission capacity, and the operating transmission current must be less than the maximum line transmission limit to ensure transmission safety: |I ij (y,u)≤I ij,max (5), where I ij I represents the current flowing through the line from node i to node j; ij,max is the maximum current amplitude that the line can carry; ④ Distributed energy output power constraint: P Gi,min ≤P i (y,u)≤P Gi,max (6), Q Gi,min ≤Q i (y,u)≤Q Gi,max (7), where P Gi,min , P Gi,max They represent the minimum and maximum values ​​of the distributed energy output active power of the i-th node respectively; Q Gi,min , Q Gi,max Output the minimum and maximum values ​​of reactive power respectively; P i (y,u),Q i (y,u) are respectively the active and reactive power output of the distributed energy of the ith node under current operation; ⑤ Wind farm dispatch power constraint: 0≤P Gi (y,u)≤P Gi,max (8), where P Gi,max represents the distributed energy active output of the i-th node; P Gi(y,u) represents the active power actually injected into the grid by the distributed energy connected to the i-th node; ⑥ Wind abandonment constraint, that is, the wind power generation power abandoned is lower than: 0≤ΔP w,t ≤P w,t (9), where ΔP w,t is the abandoned air volume; P w,t is the wind power generation power.

[0059] The flexibility resource constraints include: ① Transformer margin constraint, which means that the flexibility provided by the transformer is lower than the transformer flexibility margin, as shown in the following formula: P Tmin ≤ΔP T ≤P Tmax (10), where ΔP T Represents the active power provided by the transformer; P Tmin Represents the minimum active power that the transformer can provide; P Tmax is the maximum active power that can be provided; ② interruptible load constraint, which means that the flexibility provided by the interruptible load must be lower than the maximum flexibility of the interruptible load, as shown in the following formula: 0≤P IL ≤P ILMax (11), where P IL Represents the active power provided by the interruptible load; P ILMax Represents the maximum active power that the interruptible load can provide.

[0060] Figure 5 Figure 1 is a schematic diagram of distributed energy scenario evaluation and analysis. Based on the distribution network economic dispatch optimization model, flexibility attribution analysis is carried out. The core idea is to analyze the existence of feasible solutions for different scenarios under the above constraints. At the same time, the equality and inequality constraints (3) to (11) can be converted into equality constraints by using the relaxation variable technology. The constraint set of the economic dispatch problem of flexible resources can be finally expressed in the following abbreviated form: H(x) = 0 (12). The scenarios are screened according to the satisfaction of the constraint set (12), that is, the feasible solutions of the constraint set (12) are solved for each scenario, and the constraint satisfaction of the solution result is calculated. According to the situation, the typical output scenario can be divided into feasible scenarios and infeasible scenarios; after solving the method of fast calculation of feasible solutions based on dynamic trajectories, when the typical output scenario is infeasible under the condition of constraint set (12), the over-limit situation of the solution of the infeasible scenario in constraint set (12) is further analyzed in detail, and the constraints with the largest over-limit amount are classified to determine the type of cause leading to the insufficient flexibility of the distribution network system; then, for the feasible output scenario, the feasible scenarios close to over-limit in constraint set (12) are further analyzed, and the categories of the constraints close to over-limit are distinguished and the reasons for the potential lack of flexibility are determined;

[0061] in Figure 5 The typical scenarios represented are analyzed into three categories: feasible scenarios refer to scenarios that satisfy constraint set (12), that is, H(x) = 0; infeasible scenarios refer to scenarios that do not satisfy constraint set (12), that is, H(x) ≠ 0; critical scenarios refer to scenarios in which some constraints in the solution do not exceed the limit, but the distance from the limit value of the corresponding constraint and the ratio of the constraint range to which they belong are less than the set value (1% or 0.5%).

[0062] Figure 6 It is a 14-node structure diagram and a schematic diagram of the distributed energy access location. To illustrate the feasibility and effectiveness of the method proposed in the invention, the figure takes the IEEE 14-node distribution network with 14 nodes and 14 branches as the network basis of the example, where node 1 is a balancing node, and photovoltaic generators are connected to nodes 8 and 9 respectively; the initial voltage safety interval is set to [0.95, 1.05] pu; further, the data of photovoltaic fields numbered #07806 and #07811 provided by the Renewable Energy Laboratory are used as historical measurement data, and the output scenarios are modeled and sampled based on the Copula function and the LHSD method to generate 1000 scenario samples.

[0063] Figure 7 This is a schematic diagram of a typical scenario based on the K-means clustering algorithm; Figure 8 Schematic diagram of a typical scenario based on the hierarchical K-means clustering algorithm; the typical scenarios obtained under the two K-means clustering methods are compared as shown in the figure. Figure 7 and Figure 8 As shown, it can be seen that compared with the clustering results of the traditional K-means clustering algorithm, the typical scenarios obtained by the K-means clustering algorithm based on the hierarchical idea can better represent the correlation and uncertainty between the actual output scenarios of photovoltaic power generation.

[0064] Fig. 9 This is the maximum limit-crossing constraint analysis result diagram under the initial sampling scenario (1000); Fig.10 The maximum over-limit constraint analysis results of typical scenarios (100) after clustering are shown in Figure 2. Among them, the reasons that lead to the infeasibility of the scenario are mainly concentrated in four types of constraints, namely the active and reactive constraints of distributed energy output, voltage constraints, and line transmission limit constraints. By comparing Fig. 9 and Fig.10 It can be seen that the distribution results of the selected scenarios under typical scenarios are consistent with the results of the original simulation scenario samples, which also shows that a small number of typical scenarios obtained by the hierarchical clustering algorithm can represent the original simulation scenarios for the flexibility analysis of the distribution network;

[0065] from Fig. 9 and Fig.10From the constraint analysis of the infeasible scenarios in the formula (6), it can be seen that the reason for the lack of flexibility of the current distribution network under the optimization scheduling application is that the upper limit of the active power constraint of the distributed power output in formula (6) is too large. Therefore, under the conditions of this distribution network, a method for improving the flexibility index of the active power constraint of the distributed power output in formula (6) can be selected. Further, the critical scenarios in the feasible scenarios are screened, that is, the output scenarios with a ratio of the distance from a certain type of constraint limit to the constraint range less than or equal to 0.05 are screened and used as critical scenarios. The active power constraint of the distributed power output in the critical scenario is still the largest. In addition, the statistical number of the reactive power constraint of the distributed energy output has a large increase, followed by the voltage constraint. Therefore, according to the constraint analysis under the critical scenario, it can be seen that the flexibility control measures related to reactive power are conducive to further improving the reactive level and voltage stability of the distributed energy access to achieve the enhancement of the flexibility of the distribution network.

[0066] The present invention is a distribution network flexibility attribution analysis method based on distributed energy access. In use, the present invention comprehensively considers the volatility and correlation of distributed energy, and the flexibility of the distribution network is correlated with the distribution network optimization application. For the distribution network with distributed energy access, a clustering method based on the hierarchical idea is used to reduce the generated distributed energy output scenarios, and a distribution network economic dispatch optimization model considering flexibility is used to screen typical scenarios. Then, from the perspective of the mathematical model of the distribution network economic dispatch optimization application, based on the feasibility of the distribution network economic dispatch optimization model, typical scenario samples are evaluated and attributed. The reasons for insufficient or lack of adjustment flexibility in the distribution network application caused by the access of distributed energy are analyzed, so as to provide effective information for the selection of the distribution network flexibility improvement method; the present invention can provide specific reasons for insufficient flexibility and potential weak links for typical scenarios of distributed energy access through flexibility attribution analysis, and this information is more conducive to the operation personnel to make decisions for the control and optimization of the distribution network; the present invention has the advantages of using a clustering method to reduce scenario samples, evaluate and attribute the scenario samples, and improve the flexibility of the distribution network.

Claims

1. A distribution network flexibility attribution analysis method based on distributed energy access, characterized by: The method comprises the following steps: Step 1: A power scenario generation tool consisting of Copula modeling and LHSD sampling method, which can be used to effectively model the uncertainty of distributed energy; Step 2: Using a clustering method based on hierarchical thinking, the output scenario samples generated in step 1 are effectively reduced to obtain more representative output scenarios; Step 3: After obtaining typical scenarios through the hierarchical K-means clustering algorithm in step 2, a scheduling optimization model that takes into account flexibility resources and wind turbine access is constructed; Step 4: Conduct flexibility attribution analysis based on the distribution network economic dispatch optimization model constructed in Step 3 to understand the specific reasons for insufficient flexibility and potential weak links, and provide information for users to choose effective regulation strategies to improve flexibility.

2. A distribution network flexibility attribution analysis method based on distributed energy access as claimed in claim 1, characterized in that: The output scenario generation tool composed of Copula modeling and LHSD sampling method in step 1 can effectively model the uncertainty of distributed energy by using the output scenario generation tool, specifically: Copula modeling is a statistical modeling method used to describe the dependency relationship between random variables; in Copula modeling, marginal distribution and dependency structure are modeled separately to more accurately describe the dependency relationship between variables; LHSD sampling method is a statistical method for designing parameter space sampling, which is an extension of LHS. LHSD is usually used with Copula to generate samples consistent with the joint distribution; using them to form an output scenario generation tool to perform scenario generation, specifically including the following steps: Step 1.1: Initialization: Prepare historical measurements and data of different renewable energy sources and input them into the tool, where the marginal distribution functions are usually presented as histograms; Step 1.2: Dependency modeling: Assuming the interdependence structure, a multivariate joint distribution is constructed to represent the interdependence between renewable energy sources using linkage relations; this contains information about the degree of dependence and the dependency structure; Step 1.3: Ranking and screening: Calculate the index of goodness of fit using Copula and select the best correlation model in the sampling stage; Step 1.4: Scenario generation: Based on the multivariate joint distribution and dependency model obtained in the previous stage, LHSD is applied to generate scenarios; Step 1.5: Output Analysis: Analyze the generated scenario.

3. A distribution network flexibility attribution analysis method based on distributed energy access as claimed in claim 2, characterized in that: The step 2 uses a clustering method based on hierarchical thinking to effectively reduce the output scenario samples generated in step 1 to obtain more representative output scenarios. Specifically, the clustering algorithm based on hierarchical thinking is a variation of the K-means clustering method. This method evaluates similarity based on sample distance and takes the sum of the distances between each sample and the center point of its category as the optimization target. This method can further improve the efficiency and accuracy of the K-means method and apply it to the reduction of distributed energy output scenarios. The implementation of the K-means algorithm based on hierarchical thinking specifically includes two stages: building a hierarchical data set and clustering layer by layer.

4. The distribution network flexibility attribution analysis method based on distributed energy access according to claim 3 is characterized by: The first stage of the K-means algorithm based on the hierarchical idea in step 2 is specifically: the hierarchical data set is built "from top to bottom", and its purpose is to simplify the original data set layer by layer; the specific content of this stage is divided into two steps. First, in order to quickly reduce the size of the upper data set, half of the total number of individuals in the upper data set can be used as the number of groups in the lower layer, and then the K-means algorithm is used to perform clustering with the set number of groups; secondly, in order to ensure that the similarity between the simplified data set and the upper data set meets the requirements, the groups obtained by the K-means algorithm are screened for "out-of-group individuals", and the screening basis is to evaluate the similarity between the individuals in each group and the center, that is: Among them, p is a member in the group; c is the central individual of the group; ε is the similarity between the member and the central individual of the group; the user can set the specific similarity limit according to the type and scale of the data set. When the corresponding similarity of the member does not meet the requirements, it is judged as an "outlier individual". Then, the center of the group obtained by K-means algorithm clustering and the "outlier individual" obtained by screening together constitute the simplified data set of the lower layer. As the number of layers of the hierarchical data set increases, the similarity between the simplified data set and the original data set gradually decreases, and then the stopping condition of the hierarchical data set is constructed.

5. The distribution network flexibility attribution analysis method based on distributed energy access according to claim 3, characterized in that: The second stage of the K-means algorithm based on the hierarchical idea in step 2 is specifically as follows: the second stage uses the K-means algorithm to perform clustering layer by layer "from bottom to top"; first, from the simplified data set of the lowest layer, the initial center point is selected according to the expected number of thing groups, which is more in line with the ideal clustering requirements than the quality of the randomly selected initial center point of the traditional K-means algorithm; secondly, the new center point obtained by clustering the simplified data set of the lower layer through the K-means algorithm will be used as the initial center point of the clustering of the data set of the upper layer.

6. A distribution network flexibility attribution analysis method based on distributed energy access as claimed in claim 3, characterized in that: After obtaining a typical scenario through the hierarchical K-means clustering algorithm in step 2 in step 3, a scheduling optimization model taking into account flexibility resources and wind turbine access is constructed. Specifically, the model is based on the maximum acceptance scheduling with the minimum wind abandonment, that is, under the premise of ensuring the safety and stability of the system, the wind abandonment of wind power is minimized through optimized scheduling; at the same time, the economic efficiency of utilizing flexibility resources is considered, in which the main transformer and interruptible load of the distribution network are regarded as available flexibility resources, and the objective function of the model includes the penalty cost function of wind abandonment, the power purchase cost, and the interruptible load compensation fee; then the model of the objective function is: Where P P,t C is the amount of electricity purchased; P is the unit electricity purchase cost; P W,t is the abandoned air volume; C W P is the penalty cost per unit wind abandonment; L,t is the load shedding capacity; C L is the penalty cost per unit load shedding; P IL,t is the interruptible load reduction; C IL is the compensation cost per unit interruptible load; T represents the time period; the constraints of the optimization model include system operation constraints and flexibility resource constraints.

7. A distribution network flexibility attribution analysis method based on distributed energy access as claimed in claim 6, characterized in that: The system operation constraints of the scheduling optimization model taking into account the flexibility resources and wind turbine access in step 3 include: ① The power flow balance equation constraint refers to the calculation equation of the active and reactive power injected into the node, which can be expressed in the following compact form: f(y,u)=0(3), where y is the state variable, i.e., the voltage amplitude and phase angle of the unbalanced node; u is the control vector; ② The voltage constraint refers to the upper and lower voltage limits that the operating voltage must meet, which can be expressed as: V min ≤V i ≤V max (4), where V i represents the voltage amplitude of the ith unbalanced node; V min 、V max are the maximum and minimum values ​​of the node voltage respectively; ③ Transmission limit constraint means that each line has a certain transmission capacity, and the operating transmission current must be less than the maximum line transmission limit to ensure transmission safety: |I ij (y,u)|≤I ij,max (5), where I ij I represents the current flowing through the line from node i to node j; ij,max is the maximum current amplitude that the line can carry; ④ Distributed energy output power constraint: P Gi,min ≤P i (y,u)≤P Gi,max (6), Q Gi,min ≤Q i (y,u)≤Q Gi,max (7), where P Gi,min , P Gi,max They represent the minimum and maximum values ​​of the distributed energy output active power of the i-th node respectively; Q Gi,min , Q Gi,max Output the minimum and maximum values ​​of reactive power respectively; P i (y,u),Q i (y,u) are respectively the active and reactive power output of the distributed energy of the ith node under current operation; ⑤ Wind farm dispatch power constraint: 0≤P Gi (y,u)≤P Gi,max (8), where P Gi,max represents the distributed energy active output of the i-th node; P Gi (y,u) represents the active power actually injected into the grid by the distributed energy connected to the i-th node; ⑥ Wind abandonment constraint, that is, the wind power generation power abandoned is lower than: 0≤ΔP w,t ≤P w,t (9), where ΔP w,t is the abandoned air volume; P w,t is the wind power generation power.

8. A distribution network flexibility attribution analysis method based on distributed energy access as claimed in claim 6, characterized in that: The flexibility resource constraints of the scheduling optimization model taking into account the flexibility resources and wind turbine access in step 3 include: ① Transformer margin constraint, which means that the flexibility provided by the transformer is lower than the flexibility margin of the transformer, as shown in the following formula: P Tmin ≤ΔP T ≤P Tmax (10), where ΔP T Represents the active power provided by the transformer; P Tmin Represents the minimum active power that the transformer can provide; P Tmax is the maximum active power that can be provided; ② interruptible load constraint, which means that the flexibility provided by the interruptible load must be lower than the maximum flexibility of the interruptible load, as shown in the following formula: 0≤P IL ≤P ILMax (11), where P IL Represents the active power provided by the interruptible load; P ILMax Represents the maximum active power that the interruptible load can provide.

9. A distribution network flexibility attribution analysis method based on distributed energy access as claimed in claim 6, characterized in that: In step 4, flexibility attribution analysis is performed on the basis of the distribution network economic dispatch optimization model constructed in step 3, so as to understand the specific reasons for insufficient flexibility and potential weak links, and provide information for users to choose effective control strategies for improving flexibility. Specifically, based on the distribution network economic dispatch optimization model, flexibility attribution analysis is performed. The core idea is to analyze the existence of feasible solutions for different scenarios under the above constraints. At the same time, the equality and inequality constraints (3) to (11) can be converted into equality constraints by using the relaxation variable technology, and the constraint set of the economic dispatch problem of considering flexibility resources can be finally expressed in the following abbreviated form: H(x) = 0 (12), and the field is constrained according to the satisfaction of the constraint set (12). Scenario screening, that is, solving the feasible solutions of constraint set (12) under each scenario, and dividing the typical output scenario into feasible scenarios and infeasible scenarios according to the constraint satisfaction of the solution results; after solving the typical output scenario by the method of fast calculation of feasible solutions based on dynamic trajectories, when the typical output scenario is infeasible under the condition of constraint set (12), further detailed analysis is conducted on the over-limit situation of the solution of the infeasible scenario under constraint set (12), and the constraints with the largest over-limit amount are classified to determine the type of cause leading to the insufficient flexibility of the distribution network system; then, for the feasible output scenario, further analysis is conducted on the feasible scenarios close to over-limit in constraint set (12), and then the categories of the constraints close to over-limit are distinguished and the cause of the potential insufficient flexibility is determined; Therefore, typical scenarios are analyzed into three categories: feasible scenarios refer to scenarios that satisfy constraint set (12), that is, H(x) = 0; infeasible scenarios refer to scenarios that do not satisfy constraint set (12), that is, H(x) ≠ 0; critical scenarios refer to scenarios in which some constraints in the solution do not exceed the limit, but the distance from the limit value of the corresponding constraint and the ratio of the constraint range are less than the set value.

Images