Distributed resource distribution network control method and device based on dynamic partitioning

Through dynamic partitioning control method, combined with hierarchical analysis method and Monte Carlo-fuzzy evaluation, the distribution power output and source network load storage self-optimization operation are optimized, and the regulation complex caused by multi-faceted wide and non-deterministic coupling of distributed resource points is solved. The degree problem has achieved efficient multi-voltage grade distribution network regulation.

CN118508445BActive Publication Date: 2024-11-26STATE GRID SHANDONG ELECTRIC POWER CO +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410982121.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-26
Estimated Expiration
2044-07-22

AI Technical Summary

Technical Problem

The existing technology is difficult to effectively regulate the problem of high-proportion distributed resource points with wide surfaces, the 'explosion' of regulation dimensions caused by the coupling of distributed access and strong non-deterministic, the rapid increase in regulation complexity, excessive calculation pressure on the main station, and static partitioning The method cannot reflect the dynamic characteristics of distributed power supply and electrified load, and the research conclusion cannot meet the needs of actual medium and low voltage multi-voltage distribution networks.

Method used

The distributed resource distribution network regulation method based on dynamic partition is adopted, and the distribution network partition comprehensive evaluation system is constructed by collecting data, and the index weighting is empowered based on the combination of hierarchical analysis method-entropy weight method is adopted. The 'Monte Carlo-fuzzy evaluation' method is adopted. Generate a partitioning scheme, optimize the independent output based on the external consistency control protocol, determine the self-optimization operation strategy of the source network load storage based on the distributed consistency algorithm, and perform hierarchical coordination control.

Benefits of technology

The regulation dimension is reduced, the regulation frequency is reduced, the distribution network's ability to absorb distributed resources and clean energy is improved, the voltage limit and power reversal is solved, and the operation and regulation capabilities of the distribution network are enhanced. Meet the actual operational needs of multi-voltage levels.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118508445B_ABST
    Figure CN118508445B_ABST
Patent Text Reader

Abstract

The present invention discloses a distributed resource distribution network control method and device based on dynamic partitioning, which belongs to the technical field of distribution network control. The method includes the following steps: collecting distributed resource distribution network data and constructing a comprehensive evaluation system for distribution network partitioning; weighting indicators based on a combination of analytic hierarchy process and entropy weight method, and generating a distribution network partitioning scheme using a two-level sample generation and screening method of "Monte Carlo-Fuzzy Evaluation"; autonomously optimizing the output of distributed power sources based on an external consistency control protocol; determining the source-grid-load-storage self-optimization operation strategy of distribution network partitioning based on a distributed consistency algorithm; and hierarchically coordinating and controlling the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy. The present invention solves the problems of voltage over-limit and power reverse transmission caused by the access of distributed power sources, greatly enhances the distribution network's ability to absorb distributed resources and clean energy, and accelerates the construction process of new power systems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a distributed resource distribution network control method and device based on dynamic partitioning, belonging to the technical field of distribution network control. Background Art

[0002] Distributed energy is a type of energy supply built at the user end. It can be operated independently or in grid connection. It is a system that determines the mode and capacity based on the maximum resource and environmental benefits. It integrates and optimizes the various energy needs of users and the resource allocation status. It adopts a new energy system with demand-responsive design and modular configuration. It is a decentralized energy supply method relative to centralized energy supply. Distributed energy has the characteristics of reasonable energy efficiency, low loss, low pollution, flexible operation, and good system economy. However, the new development situation of distributed clean energy has put forward higher requirements for the improvement of the carrying capacity, access capacity, and operation control capacity of the distribution network, and has also brought a series of severe challenges and difficulties, as follows: the large-scale access of distributed power sources with many points and wide areas and impact loads such as electric vehicles and rail transit has significantly deepened the degree of power electronics in the distribution system, increased uncertainty, and made the bidirectional power flow characteristics more prominent. The AC / DC hybrid form has been continuously evolving, and the electrical coupling and energy information coupling between "source-grid-load-storage" have become closer. It is an inevitable trend for the distribution system to develop into a new type of distribution system with a high proportion of distributed power sources, a high proportion of power electronic equipment / load access, high uncertainty and strong randomness, and explosive growth of massive data.

[0003] The emergence of new features of new distribution systems has significantly increased the complexity of the control system and reduced the safety and reliability of the system, and will even bring revolutionary changes to the planning, design, operation control, fault protection and other aspects of the distribution system. Researchers in related fields have conducted extensive research on the above issues, mainly focusing on distribution network zoning methods (the basis of group control and group regulation) and coordinated control methods based on hierarchical zoning. The basic technical route is to reduce the control dimension through zoning and decouple the electrical coupling between different voltage levels through stratification. Some research progress has been made as follows: (1) In the distribution network zoning dimension, some researchers use electrical distance as the basis for judgment to characterize the electrical coupling tightness between nodes in the partition, and divide nodes with higher coupling tightness into the same partition; some researchers integrate voltage sensitivity and voltage safety domain into the judgment index of electrical distance. This regional division method is a zoning method that focuses on voltage safety based on the electrical coupling tightness. Some researchers further considered the needs of local consumption of distributed power sources and autonomy, tight coupling within the partition, and weak coupling between partitions, and proposed a distributed resource regional division method. (2) In terms of hierarchical and zoned control of distribution networks, most studies focus on distributed power control in low-voltage areas, with local control as the main control method, such as local control methods based on voltage-reactive power control and droop control of distributed power inverters. Low-voltage local control is usually prone to fall into local optimality. Relevant researchers have further conducted research on low-voltage distributed control and medium-low voltage hierarchical coordinated control, but such research is still relatively scarce.

[0004] Although the above research results provide a certain degree of theoretical and methodological support for the operation and control of new distribution systems with large-scale distributed resource access, most of them are still at the level of scientific research, and fail to consider the problems and needs that may be faced on site during actual application, and fail to achieve an effective combination of research, learning and application. Specifically, there are the following problems: (1) In the dimension of regional division, the main problem of current research is that the elements involved in most studies are not comprehensive, focusing only on one of the distributed power sources or loads, and failing to better reflect the interactive characteristics between sources and loads; secondly, the static partitioning method accounts for a high proportion, that is, the average value of the distributed power output or load demand over a period of time is used as the calculation basis to obtain a partitioning method within a longer time scale. This partitioning method cannot reflect the dynamic characteristics of distributed power sources and electrified loads affected by weather and user habits. The static partitioning method is not very meaningful; finally, a considerable number of the partitioning strategies proposed by the research are only tested and verified through single voltage level examples (such as IEEE-33, PG&E-69 examples, etc.), and the research conclusions obtained cannot meet the actual regulation needs of medium and low voltage multi-voltage level distribution networks. (2) In terms of hierarchical and zoned control, the main problem with current research is that a considerable number of studies are based on single-voltage distribution networks, ignoring the energy interaction between different voltage levels; secondly, the data sources required for control are overly idealized, resulting in the generated control strategies being inconsistent with current on-site actual applications and difficult to apply. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a distributed resource distribution network control method and device based on dynamic partitioning, which can solve the actual pain points of "explosion" of control dimensions, sharp increase in control complexity, and excessive computing pressure on the master station caused by the extremely high proportion of distributed resource points, widespread and decentralized access and strong non-deterministic coupling in actual production.

[0006] The technical solution adopted by the present invention to solve the technical problem is:

[0007] In a first aspect, an embodiment of the present invention provides a distributed resource distribution network control method based on dynamic partitioning, comprising the following steps:

[0008] Step 1: Collect distributed resource distribution network data and build a comprehensive evaluation system for distribution network partitions, where the comprehensive evaluation system includes a source-load matching index, a network loss index, a voltage deviation index, and an electrical distance index;

[0009] Step 2: weight the indicators based on the combination of analytic hierarchy process and entropy weight method, and use the two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation" to generate the distribution network partitioning scheme;

[0010] Step 3: autonomously optimize the output of distributed power sources based on the external consistency control protocol;

[0011] Step 4, determining the source-grid-load-storage self-optimization operation strategy of the distribution network partition based on the distributed consistency algorithm;

[0012] Step 5: Perform hierarchical coordinated control of the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy.

[0013] As a possible implementation of this embodiment, the source-load matching index includes the new energy consumption rate of the distribution network when participating in AGC and AVC scheduling:

[0014] (1)

[0015] (2)

[0016] in, D 1 is the new energy consumption rate when the distribution network participates in AGC dispatching, D 2 is the new energy consumption rate when the distribution network participates in AVC dispatch, B is the set of nodes, is the set of distributed renewable energy in node i, and They represent the active output of the flexible resource l at time t when the distribution network responds to AGC and AVC instructions, respectively. represents the maximum active power of flexible resource l at time t;

[0017] The network loss index includes a first network loss index D 3 and second network loss index D 4:

[0018] (3)

[0019] (4)

[0020] in, r ( i ) is a node i The collection of distributed resources in (including distributed new energy and other distributed resources), p i D express i The active power of the node load, p PCC,t express t The active power injected into the grid by the PCC point at the moment;

[0021] The voltage deviation index includes a first voltage deviation rate index D 5 and the second voltage deviation rate indicator D 6:

[0022] (5)

[0023] (6)

[0024] in, V i,t AGC and V i,t AVC They represent the voltage of node i at time t when the distribution network participates in AGC and AVC regulation, V B Indicates the voltage reference value, usually the rated voltage of the PCC point;

[0025] The electrical distance indicator includes the average electrical distance d of the partition:

[0026] (7)

[0027] in, γ (m) indicates partition m The set of nodes in dd ij Representation Node j The ratio of the voltage change value of the node itself to the voltage change value of node i when a unit reactive power change occurs:

[0028] (8)

[0029] in, The voltage-reactive sensitivity matrix is ​​represented by i Line j List.

[0030] As a possible implementation of this embodiment, step 2, weighting indicators based on a combination of analytic hierarchy process and entropy weight method, and using a two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation" to generate a distribution network partitioning scheme, includes:

[0031] Based on the combination of analytic hierarchy process and entropy weight method, the weighting and level division of zoning indicators are carried out;

[0032] The "Monte Carlo-Fuzzy Evaluation" two-level sample generation and screening method is used to generate a set of distribution network partitioning schemes under different working conditions;

[0033] The distribution network partitioning scheme set is evaluated and optimized to generate the optimal partitioning scheme.

[0034] As a possible implementation of this embodiment, the weighting and level division of partition indicators based on the combination of hierarchical analysis method and entropy weight method includes:

[0035] Defining Metrics jj The subjective weight of j for:

[0036] (9)

[0037] in, ii and jj Indicates the index number; I indicates the set of indicators; A jj,ii is the matrix A jj Line ii Column elements, whose meaning is index jj Relative to the indicator ii The importance of A jj,ii The value of is defined as follows: 1 represents the indicator jj Ratio Index ii Important, 0.5 indicates an indicator jj and indicators ii Equally important, 0 means indicator ii Ratio Index jj important;

[0038] Defining Metrics jj The objective weight of for:

[0039] (10)

[0040] in, J Indicates the number of indicators; Indicators jj The information entropy of is calculated as follows:

[0041] (11)

[0042] in, N Indicates the total number of samples participating in the evaluation. Representation sample n For jj The probability of the evaluation level of an indicator is expressed as:

[0043] (12)

[0044] in, Representation sample n Medium Index jj The evaluation level of

[0045] The subjective weight and the objective weight are multiplied and combined to obtain the index. jj The combined weight of for:

[0046] (13);

[0047] The indicators are divided into four levels: excellent, good, ordinary and poor, with values ​​of 1, 0.75, 0.5 and 1 respectively. The dividing lines line1~line3 between the four levels are determined by maximizing the numerical differences between the four levels.

[0048] As a possible implementation of this embodiment, the two-level sample generation and screening method of "Monte Carlo-Fuzzy Evaluation" is used to generate a set of distribution network partitioning schemes under different working conditions, including:

[0049] Generate partitioned random samples based on the Monte Carlo method;

[0050] The fuzzy evaluation method is used to screen the large number of generated samples to obtain partition samples that are more in line with the actual scenario.

[0051] As a possible implementation of this embodiment, the step of evaluating and optimizing the distribution network partitioning scheme set to generate an optimal partitioning scheme includes:

[0052] Preliminary evaluation of samples: First, determine the level dividing lines line1~line3 of each indicator; then generate the indicator-level probability matrix of each sample; then perform combined weighting of indicators; finally, use fuzzy evaluation method to grade and evaluate the samples, and obtain the overall fuzzy evaluation level of each sample;

[0053] Sample screening and secondary evaluation: First, select excellent samples and re-divide the grade boundaries line1~line3 for excellent samples; then, use the new grade boundaries to re-evaluate the indicators of each excellent sample to obtain a new evaluation level for each sample, and use the entropy weight method to calculate the corresponding objective weight; finally, combine the subjective weight with the new objective weight to obtain the new combined weight when the flexible resource cluster participates in AGC and AVC regulation, and finally obtain the comprehensive evaluation score of each sample participating in AGC and AVC, that is:

[0054] (19)

[0055] (20)

[0056] in, and , sample n Medium Index jj The new rating level, Respectively represent the indicators when the flexible resource cluster participates in AGC and AVC regulation jj New portfolio weights;

[0057] Result-oriented weight optimization and correction: The evaluation results of theoretical calculations are verified for correlation with the actual scheduling effects. If the two show a low correlation, a sequential arrangement matrix of the partition method is constructed based on the actual scheduling effect, and the weight of each indicator is optimized and corrected using the "Monte Carlo-Fuzzy Evaluation" two-level sample generation and screening method.

[0058] As a possible implementation method of this embodiment, the autonomous output optimization of distributed power sources based on the external consistency control protocol includes: in a distribution network containing multiple DGs, firstly, the distributed power sources connected to the distribution network system are grouped according to different control targets, and are divided into voltage control groups and power control groups; then, the leading DG and follower DG in the group and the consistency variables are selected, and the leading DG detects the information of the control targets of the group; then, based on the external consistency protocol, the state consistency of the distributed power sources in the group, the voltage regulation of the distribution network nodes and the power control at the public connection are achieved, so as to realize the self-optimizing operation at the device level.

[0059] As a possible implementation of this embodiment, the source-grid-load-storage self-optimization operation strategy for determining the distribution network partition based on the distributed consistency algorithm includes:

[0060] Establish optimization model and analyze the cost and benefit functions of each part;

[0061] The Lagrange multiplier method is used to solve the optimization model and obtain the final optimization model.

[0062] As a possible implementation of this embodiment, the optimization model establishes an objective function with the best economic performance of the subnet as the goal:

[0063] (twenty one),

[0064] In the formula, is the total benefit of the autonomous subnet, They represent the collection of all temperature-controlled flexible loads, non-temperature-controlled flexible loads, new energy units, and battery energy storage units in the autonomous area; , They represent the benefits obtained by the optimal scheduling of the temperature control flexible load and the non-temperature control flexible load participating in the subnet; represents the operating cost of battery energy storage; represents the total cost of photovoltaics; The set temperature for temperature control loads; The power demand of non-temperature-controlled loads; Output active power for photovoltaic; The active power output or stored by the energy storage unit;

[0065] The cost and benefit functions of each part of the analysis include:

[0066] Benefit function of temperature control flexible load:

[0067] (twenty two),

[0068] In the formula, It is a coefficient used to characterize the benefit of temperature control flexible load. is the outdoor temperature, It is the temperature when the user feels the most comfortable, usually 25℃;

[0069] The electricity benefit function of non-temperature-controlled flexible loads:

[0070] (twenty three),

[0071] In the formula, They are k The coefficients of constant, linear and quadratic terms in the benefit function of non-temperature-controlled flexible loads;

[0072] The operating cost function of photovoltaic itself:

[0073] (24-1),

[0074] Photovoltaic abandonment cost function:

[0075] (24-2),

[0076] The total cost function of new energy units:

[0077] (25),

[0078] In the formula, is the cost coefficient of photovoltaic, wind and solar power abandonment; is the actual output power; is the maximum value of the output power, that is, the output power of the unit when operating in MPPT mode;

[0079] Battery energy storage cost function:

[0080] (26),

[0081] In the formula, is the coefficient of the cost function of the nth battery storage;

[0082] The operating constraints of each resource are:

[0083] (27)

[0084] Power balance constraints in each area:

[0085] (28)

[0086] In the formula, Indicates the dispatching instructions issued by the superior dispatching center; It represents the set of common loads in the area except flexible loads; represents the required power of the oth common load; is the power consumed by the jth temperature-controlled flexible load. There is a quadratic function relationship between the power and the set temperature. The coefficient of the quadratic function is ; Photovoltaic output The upper and lower limits of It is the energy storage that emits and absorbs power The upper and lower limits of The power consumption of temperature-controlled flexible load The upper and lower limits of It is the temperature setting temperature of the temperature control type flexible load The upper and lower limits of Power consumption of non-temperature-controlled flexible load The upper and lower limits of .

[0087] As a possible implementation of this embodiment, the Lagrange multiplier method is used to solve the optimization model to obtain the final optimization model, including:

[0088] Considering the inequality constraints in equation (26), the Lagrange multiplier method is applied to the optimization model, and the original optimization objective function is transformed into:

[0089] (29),

[0090] In the pair and Lagrange multipliers By taking partial derivatives, we can obtain the optimal conditions for the following equivalent optimization problem:

[0091] (30),

[0092] According to the above optimality equation, we get:

[0093] (31),

[0094] Assuming that all power components in the subnet operate within their inequality constraints, the consistency variables of various power components are defined as:

[0095] (32),

[0096] The update formula of the consistency variable of the slave node is:

[0097] (33),

[0098] The update formula of the master node is:

[0099] (34),

[0100] Where M represents the type of regulation resource, is the convergence coefficient, which is closely related to the convergence speed of the consistency variable of the master node; ΔP represents the difference between the total active power demand of all types of loads in the subnet and the active power output of all types in the current subnet;

[0101] Associate the constrained quantity with the consistency variable of the component, and satisfy the constraint during the iteration of the consistency variable:

[0102] (35),

[0103] The temperature constraint and upper and lower limit constraints of power consumption of temperature control flexible load are modified as follows:

[0104] (36),

[0105] The upper and lower power limits of non-temperature-controlled flexible loads are modified to:

[0106] (37),

[0107] The upper and lower limits of photovoltaic output are modified as follows:

[0108] (38),

[0109] The output and storage upper limit constraints of the battery energy storage unit are modified to:

[0110] (39).

[0111] As a possible implementation of this embodiment, step 5, performing hierarchical coordinated control on the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy, includes:

[0112] Based on the cloud-edge collaborative communication architecture, a new partitioned and layered distribution system control architecture with "cloud-partition-end" collaboration is established;

[0113] Based on the new zoned and layered distribution system control architecture, hierarchical coordinated control is carried out by adjusting the objective function according to the source-grid-load-storage self-optimization operation strategy.

[0114] As a possible implementation of this embodiment, the "end" of the partitioned and layered new distribution system control architecture is responsible for establishing a scheduling model for various distributed flexible resources, as well as data collection and sensing, and communicating real-time operation information with the upper layer, and automatically receiving and responding to the control instructions issued by the "partition" level;

[0115] The "partition" of the partitioned and layered new distribution system control architecture is composed of the control logic layer of each resource cluster, which is responsible for online aggregation of various resources in the cluster, obtaining the cluster aggregation model and uploading it to the cloud management and control platform; at the same time, it responds online to the control instructions issued by the cloud control center and decomposes them into various distributed resources in the group;

[0116] The "cloud" of the new partitioned and layered distribution system control architecture is a cloud control center, which generates cloud resource clusters through online dynamic clustering; collects aggregation models of each cluster and coordinates the regulation capabilities of direct-control resources in different partitions and power grids; and decomposes the dispatching instructions issued by the power grid through cloud command decomposition and sends them to each resource cluster.

[0117] As a possible implementation of this embodiment, the control objective function includes a low network active power loss objective function, a voltage deviation reduction objective function, and an improvement in the photovoltaic power generation consumption objective function in the region.

[0118] Low network active power loss objective function:

[0119] (40),

[0120] Where: N T Indicates the total number of time periods optimized; Represents the collection of all branches in the power distribution system; r ij For branch ij The resistance value; I ij,t For branch ij exist t The current flowing through it at any moment;

[0121] Reduce voltage deviation objective function:

[0122] (41),

[0123] Where: N N Represents the total number of nodes in the power distribution system; U i,t For Node i exist t The voltage amplitude at the moment;

[0124] Improve the objective function of photovoltaic power generation consumption in the region:

[0125] (42),

[0126] In the formula, N PV is the total number of photovoltaic generators in the distribution system; P i,t PV For the i Generator sets in t Actual output at the moment; P i,t PV,max For the i Generator sets in t Maximum output at any time;

[0127] According to the actual needs of distribution network operation optimization, the above optimization objectives are comprehensively considered and the optimization objective function is expressed by weight coefficients:

[0128] (43),

[0129] Where: Represents the weight coefficient, satisfying ; Represents the objective function as shown in equations (40)~(42).

[0130] In a second aspect, an embodiment of the present invention provides a distributed resource distribution network control device based on dynamic partitioning, comprising:

[0131] An evaluation system construction module is used to collect distributed resource distribution network data and construct a distribution network partition comprehensive evaluation system, wherein the distribution network partition comprehensive evaluation system includes a source-load matching index, a network loss index, a voltage deviation index, and an electrical distance index;

[0132] The distribution network partitioning module is used to assign weights to indicators based on the combination of hierarchical analysis method and entropy weight method, and to generate distribution network partitioning schemes using the two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation";

[0133] Autonomous output optimization module, used to optimize the autonomous output of distributed power sources based on external consistency control protocol;

[0134] The self-optimizing operation module is used to determine the source-grid-load-storage self-optimizing operation strategy of the distribution network partition based on the distributed consistency algorithm;

[0135] The hierarchical coordination control module is used to perform hierarchical coordination control on the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy.

[0136] In a third aspect, an embodiment of the present invention provides a computer device, comprising a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory through the bus, and the processor executes the machine-readable instructions to perform any step of the above-mentioned distributed resource distribution network control method based on dynamic partitioning.

[0137] In a fourth aspect, an embodiment of the present invention provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned distributed resource distribution network control methods based on dynamic partitioning are executed.

[0138] The beneficial effects of the technical solution of the embodiment of the present invention are as follows:

[0139] Under the premise of considering the current property rights attributes of distributed power sources and the willingness of users to participate in regulation, the present invention solves the problems of voltage over-limit and power reverse transmission caused by the access of distributed power sources. At the same time, it breaks through the practical problems of high regulation complexity, over-regulation, and poor global optimality of regulation results in the existing active distribution network regulation strategy, greatly enhances the distribution network's ability to absorb distributed resources and clean energy, and accelerates the construction process of new power systems.

[0140] The present invention adopts the technologies of "zoning first - optimizing later" and "optimizing operation within the region - coordinated control between regions", which reduces the control dimensions and the control frequency; it solves the actual pain points of "explosion" of control dimensions, sharp increase in control complexity, and excessive computing pressure on the master station caused by the extremely high proportion of distributed resource points, the wide range of decentralized access and strong non-deterministic coupling in actual production, and provides distribution network operators with an operation and control method for a new type of distribution system with multiple voltage levels that conforms to the actual operation management caliber, covers multiple voltage levels, meets a certain degree of economy, and is operational. BRIEF DESCRIPTION OF THE DRAWINGS

[0141] Figure 1 is a flow chart of a distributed resource distribution network control method based on dynamic partitioning according to an exemplary embodiment;

[0142] Figure 2 is a schematic diagram of a partition sample generation process according to an exemplary embodiment;

[0143] Figure 3 is a schematic diagram of grouping a distribution network including multiple DGs according to an exemplary embodiment;

[0144] Figure 4 is a block diagram of a distributed power supply control without external input according to an exemplary embodiment;

[0145] Figure 5 is a block diagram of a distributed power supply control with external input according to an exemplary embodiment;

[0146] Figure 6 is a schematic diagram of a "cloud-region-end" multi-level coordinated control architecture according to an exemplary embodiment;

[0147] Figure 7 It is a structural diagram of a distributed resource distribution network control device based on dynamic partitioning according to an exemplary embodiment. DETAILED DESCRIPTION

[0148] In order to more clearly illustrate the technical features of the solution of the present invention, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0149] like Figure 1 As shown, a distributed resource distribution network control method based on dynamic partitioning provided by an embodiment of the present invention includes the following steps:

[0150] Step 1: Collect distributed resource distribution network data and build a comprehensive evaluation system for distribution network partitions, where the comprehensive evaluation system includes a source-load matching index, a network loss index, a voltage deviation index, and an electrical distance index;

[0151] Step 2: weight the indicators based on the combination of analytic hierarchy process and entropy weight method, and use the two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation" to generate the distribution network partitioning scheme;

[0152] Step 3: autonomously optimize the output of distributed power sources based on the external consistency control protocol;

[0153] Step 4, determining the source-grid-load-storage self-optimization operation strategy of the distribution network partition based on the distributed consistency algorithm;

[0154] Step 5: Perform hierarchical coordinated control of the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy.

[0155] As a possible implementation of this embodiment, the distribution network zoning comprehensive evaluation system covers 4 categories of indicators for a total of 7 evaluation area division methods. In order to measure the amount of wind and solar power curtailment within the zone, the source-load matching index defines the new energy consumption rate of the distribution network when participating in AGC and AVC scheduling:

[0156] (1)

[0157] (2)

[0158] in, D1 is the new energy consumption rate when the distribution network participates in AGC dispatching, D 2 is the new energy consumption rate when the distribution network participates in AVC dispatch, B is the set of nodes, is the set of distributed renewable energy in node i, and They represent the active output of the flexible resource l at time t when the distribution network responds to AGC and AVC instructions, respectively. represents the maximum active power of flexible resource l at time t; from equations (1) and (2), we can see that D 1 and D 2 respectively represent the time periods T The value is the average ratio of the output of all distributed renewable energy sources to their maximum available output when participating in AGC and AVC regulation. The larger the value is, the higher the renewable energy consumption rate will be when the distribution network responds to AGC and AVC dispatching instructions.

[0159] When the distribution network participates in the dispatch of the upper power grid, although the power of the PCC point can track the dispatch instruction to a certain extent under various regional division methods, the output of the same flexible resource may be different under different partitioning methods, resulting in different losses in the distribution network. In order to measure the loss of the distribution network when participating in AGC and AVC dispatch under different partitioning methods, the network loss index defines the first network loss index D 3 and second network loss index D 4:

[0160] (3)

[0161] (4)

[0162] in, r ( i ) is a node i The collection of distributed resources in (including distributed new energy and other distributed resources), p i D express i The active power of the node load, p PCC,t express t The active power injected into the grid by the PCC point at the moment; From equations (3) and (4), we can know that D 3 and D 4 The brackets contain three items. The first item is the sum of the output of all flexible resources participating in AGC and AVC control at the current moment. The second item represents the load active power. The third item is the active power injected by the upper power grid. The sum of items 1 and 3 is the total active power input to the distribution network. After deducting the load active power of item 2, it is the network loss. A good zoning method is to minimize the regional D3 and D 4.

[0163] Regardless of whether the adjustable resources of the distribution network participate in regulation in a single point form or in various partition forms, under the premise of meeting basic constraints, a smaller voltage offset will help improve the stability of the distribution network participating in the regulation of the upper power grid. In order to measure the overall offset of the node voltage when the distribution network participates in AGC and AVC regulation, the voltage deviation index defines the first voltage offset rate index D 5 and the second voltage deviation rate indicator D 6:

[0164] (5)

[0165] (6)

[0166] in, V i,t AGC and V i,t AVC They represent the voltage of node i at time t when the distribution network participates in AGC and AVC regulation, V B represents the voltage reference value, which is usually the rated voltage of the PCC point. From equations (5) and (6), we can see that: D 5 and D 6 is the time period T The average value of the sum of the deviations between the per-unit voltage of each node and the base voltage. Obviously, the larger the value, the greater the overall deviation of the grid voltage and the worse the regulation performance.

[0167] The degree of coupling between nodes in a partition is related to the difficulty of internal regulation of the aggregator. Considering the balance of interests between the power grid and the aggregator, the electrical distance indicator defines the average electrical distance d of the partition:

[0168] (7)

[0169] in, γ (m) indicates partition m The set of nodes in dd ij Representation Node j The ratio of the voltage change value of the node itself to the voltage change value of node i when a unit reactive power change occurs:

[0170] (8)

[0171] in, The voltage-reactive sensitivity matrix is ​​represented by i Line j List; ddij The larger the node j With Node i The looser the coupling relationship, the greater the electrical distance.

[0172] As a possible implementation of this embodiment, step 2, weighting indicators based on a combination of analytic hierarchy process and entropy weight method, and using a two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation" to generate a distribution network partitioning scheme, includes:

[0173] Based on the combination of analytic hierarchy process and entropy weight method, the weighting and level division of zoning indicators are carried out;

[0174] The "Monte Carlo-Fuzzy Evaluation" two-level sample generation and screening method is used to generate a set of distribution network partitioning schemes under different working conditions;

[0175] The distribution network partitioning scheme set is evaluated and optimized to generate the optimal partitioning scheme.

[0176] As a possible implementation of this embodiment, the weighting and level division of partition indicators based on the combination of hierarchical analysis method and entropy weight method includes:

[0177] Defining Metrics jj The subjective weight of j for:

[0178] (9)

[0179] in, ii and jj Indicates the index number; I indicates the set of indicators; A jj,ii is the matrix A jj Line ii Column elements, whose meaning is index jj Relative to the indicator ii The importance of A jj,ii The value of is defined as shown in Table 1.

[0180] Table 1: A jj,ii Definition

[0181]

[0182] As an objective weighting method, the entropy weight method determines the amount of information provided by a certain indicator by calculating the information entropy, highlighting the difference in the amount of information provided by the indicator. Based on this idea, the indicator is defined jj The objective weight of for:

[0183] (10)

[0184] in, J Indicates the number of indicators; Indicators jj The information entropy of is calculated as follows:

[0185] (11)

[0186] in, N Indicates the total number of samples participating in the evaluation. Representation sample n For jj The probability of the evaluation level of an indicator is expressed as:

[0187] (12)

[0188] in, Representation sample n Medium Index jj The specific values ​​of the evaluation level are shown in Table 2 (a total of 4 levels are defined, namely excellent, good, ordinary and poor).

[0189] Table 2: R n,jj Definition

[0190]

[0191] The subjective weight and the objective weight are multiplied and combined to obtain the index. jj The combined weight of for:

[0192] (13);

[0193] The indicators are divided into four levels: excellent, good, ordinary and poor, with values ​​of 1, 0.75, 0.5 and 1 respectively. The dividing lines line1~line3 between the four levels are defined according to Table 2. The dividing lines line1~line3 between the four levels are determined by maximizing the numerical differences between the four levels.

[0194] As a possible implementation of this embodiment, the specific process of determining the dividing lines line1 to line3 between the four levels by maximizing the numerical differences between the four levels is as follows:

[0195] 1) Define the difference function ∆( kk ) is the level kk The average distance between all data and their mean:

[0196] (14)

[0197] Among them, U kk Indicates the corresponding indicator kk Level collection, X kk express kk The number of samples in the class, express x The mean of

[0198] 2) Based on the difference function ∆( kk ), with the goal of minimizing the difference between data at the same level or maximizing the difference between data at different levels, construct the objective function and constraints:

[0199] (15)

[0200] (16)

[0201] 3) Solve the three dividing lines of the optimization problem.

[0202] Due to the limitation of the "four capabilities" of the current low-voltage distribution network, it is impossible to generate sufficient data to generate partition instance samples. The "four capabilities" are "adjustable, controllable, observable, and measurable". The present invention proposes a "Monte Carlo-fuzzy evaluation" two-level sample generation and screening method to generate a set of distribution network partition schemes under different working conditions.

[0203] As a possible implementation of this embodiment, the two-level sample generation and screening method of "Monte Carlo-Fuzzy Evaluation" is used to generate a set of distribution network partitioning schemes under different working conditions, including:

[0204] Generate partitioned random samples based on the Monte Carlo method; Figure 2 As shown, the generation of random samples of partitions based on the Monte Carlo method includes: node 1 is connected to the upstream power grid, first, starting from node 2, all nodes are assigned partition numbers in turn to indicate which partition the node belongs to; then, it is determined whether this static partitioning method meets the static partitioning conditions, and this partitioning method is recorded if the conditions are met; finally, when the iteration time exceeds the set value, that is, when a new partitioning method that meets the conditions cannot be found, all partitioning methods are output; the static partitioning conditions include: ① the nodes inside the partition are interconnected; ② each partition has at least two resources; ③ after the partitioning is completed, all resources are in the partition; ④ there is no duplication with the existing partitioning scheme; ⑤ the average electrical distance of the partition is less than 0.2.

[0205] The fuzzy evaluation method is a decision analysis method based on fuzzy logic theory, which is used to deal with uncertainty and fuzziness. The fuzzy evaluation method is used to screen a large number of generated samples to obtain partition samples that are more in line with the actual scene; the fuzzy evaluation method is used to screen a large number of generated samples to obtain partition samples that are more in line with the actual scene, including:

[0206] 1) Determine the evaluation index of the static partitioning method and the value of each index in each partitioning sample;

[0207] 2) Adopt the combination of analytic hierarchy process and entropy weight method to determine the weight ω of the evaluation index;

[0208] 3) Construct a membership function for the comment set (using the idea of ​​index grading);

[0209] 4) Construct the fuzzy relationship matrix PR, as shown in formula (17):

[0210] (17)

[0211] in, , , , They represent the probabilities that indicator i is excellent, good, average, and poor respectively;

[0212] 5) Use appropriate synthesis factors to perform fuzzy transformation and obtain the evaluation result XR, as shown in formula (18):

[0213] (18)

[0214] Wherein, * represents a synthesis factor, and the synthesis factor of the present invention adopts multiplication;

[0215] 6) Filter out the partitions at the excellent level to form an excellent partition group.

[0216] The evaluation result XR represents the overall evaluation result of the partitioned samples. Like the four levels of the indicators, the evaluation results of the partitioned samples are also divided into four levels: excellent, good, ordinary and poor. The present invention only selects the representative excellent level partitions in the evaluation results.

[0217] As a possible implementation of this embodiment, the step of evaluating and optimizing the distribution network partitioning scheme set to generate an optimal partitioning scheme includes:

[0218] Preliminary evaluation of samples: First, determine the level dividing lines line1~line3 of each indicator; then generate the indicator-level probability matrix of each sample; then perform combined weighting of indicators; finally, use fuzzy evaluation method to grade and evaluate the samples, and obtain the overall fuzzy evaluation level of each sample;

[0219] Sample screening and secondary evaluation: First, select excellent samples and re-divide the grade boundaries line1~line3 for excellent samples; then, use the new grade boundaries to re-evaluate the indicators of each excellent sample to obtain a new evaluation level for each sample, and use the entropy weight method to calculate the corresponding objective weight; finally, combine the subjective weight with the new objective weight to obtain the new combined weight when the flexible resource cluster participates in AGC and AVC regulation, and finally obtain the comprehensive evaluation score of each sample participating in AGC and AVC, that is:

[0220] (19)

[0221] (20)

[0222] in, and They represent the comprehensive evaluation scores of sample n corresponding to AGC and AVC, respectively. sample n Medium Index jj The new rating level, and Respectively represent the indicators when the flexible resource cluster participates in AGC and AVC regulation jj New portfolio weights;

[0223] Result-oriented weight optimization and correction: The evaluation results of theoretical calculations are verified for correlation with the actual scheduling effects. If the two show a low correlation, a sequential arrangement matrix of the partitioning method is constructed based on the actual scheduling effect, and the "Monte Carlo-fuzzy evaluation" two-level sample generation and screening method is used to optimize and correct the weights of each indicator, which significantly improves the accuracy and global optimality of the final partitioning scheme.

[0224] For distributed power sources, especially those with low voltage access, the communication conditions are poor, or there is a certain communication delay (or even no communication). The system with time delay will produce complex dynamic behavior. The existence of network delay not only makes system design and consistency analysis difficult, but also makes the control performance of the system worse, and even makes the system unstable. Therefore, it is of great significance to study effective methods to reduce the impact of distributed power output on the distribution network system under disturbance. Combined with the current development trend of distributed power distribution network control technology of "group control and group adjustment", this invention first proposes a distributed power autonomous output optimization method based on external consistency control.

[0225] As a possible implementation of this embodiment, the autonomous output optimization of distributed power sources based on the external consistency control protocol includes: in a distribution network containing multiple DGs, firstly, the distributed power sources connected to the distribution network system are grouped according to different control targets, and divided into voltage control groups and power control groups, such as Figure 3 As shown; then the leader DG and follower DG in the group and the consistency variables are selected, and the leader DG detects the information of the control target of the group; then based on the external consistency protocol, the state consistency of the distributed power sources in the group, the voltage regulation of the distribution network nodes and the power control at the public connection are achieved, and the self-optimizing operation at the device level is realized. Figure 4 and Figure 5 The control block diagrams of distributed generation i designed with the consistency protocol are shown in the following figure, respectively, without external input (following DG) and with external input (leading DG). Assume that there are N distributed generation in the distribution network, M distributed generation has external input, and NM distributed generation has no external input.

[0226] In the region, system-level self-optimizing operation is implemented. The present invention first proposes a source-grid-load-storage self-optimizing operation strategy based on a distributed consistency algorithm.

[0227] As a possible implementation of this embodiment, the source-grid-load-storage self-optimization operation strategy for determining the distribution network partition based on the distributed consistency algorithm includes:

[0228] Establish optimization model and analyze the cost and benefit functions of each part;

[0229] The Lagrange multiplier method is used to solve the optimization model and obtain the final optimization model.

[0230] As a possible implementation of this embodiment, the optimization model establishes an objective function with the best economic performance of the subnet as the goal:

[0231] (twenty one),

[0232] In the formula, is the total benefit of the autonomous subnet, They represent the collection of all temperature-controlled flexible loads, non-temperature-controlled flexible loads, new energy units, and battery energy storage units in the autonomous area; , They represent the benefits obtained by the optimal scheduling of the temperature control flexible load and the non-temperature control flexible load participating in the subnet; represents the operating cost of battery energy storage; represents the total cost of photovoltaics; The set temperature for temperature control loads; The power demand of non-temperature-controlled loads; Output active power for photovoltaic; The active power output or stored by the energy storage unit;

[0233] The cost and benefit functions of each part of the analysis include:

[0234] The set temperature of temperature-controlled flexible loads (such as air conditioners) is adjustable within a certain range, which provides space for regional optimization scheduling. Since the human body's comfort feeling is different within a temperature range, the benefit of the proposed temperature-controlled load is related to the human body's comfort. The benefit is highest when the user's comfort is the highest, and the benefit is zero when the set temperature is the same as the outdoor temperature. In this way, the benefit function of the temperature-controlled flexible load is expressed by the following formula:

[0235] (twenty two),

[0236] In the formula, It is a coefficient used to characterize the benefit of temperature control flexible load. is the outdoor temperature, It is the temperature when the user feels the most comfortable, usually 25℃;

[0237] The electricity benefit function of non-temperature-controlled flexible loads can be expressed by a quadratic function:

[0238] (twenty three),

[0239] In the formula, They are k The coefficients of constant, linear and quadratic terms in the benefit function of non-temperature-controlled flexible loads;

[0240] The operating cost function of photovoltaic itself:

[0241] (24-1),

[0242] Photovoltaic abandonment cost function:

[0243] (24-2),

[0244] The total cost function of new energy units:

[0245] (25),

[0246] In the formula, is the cost coefficient of photovoltaic, wind and solar power abandonment; is the actual output power; is the maximum value of the output power, that is, the output power of the unit when operating in MPPT mode;

[0247] The battery energy storage unit can both discharge into the grid and absorb energy from the grid. The battery energy storage cost function is:

[0248] (26),

[0249] In the formula, is the coefficient of the cost function of the nth battery storage;

[0250] The operating constraints of each resource are:

[0251] (27)

[0252] Power balance constraints in each area:

[0253] (28)

[0254] In the formula, Indicates the dispatching instructions issued by the superior dispatching center; It represents the set of common loads in the area except flexible loads; represents the required power of the oth common load; is the power consumed by the jth temperature-controlled flexible load. There is a quadratic function relationship between the power and the set temperature. The coefficient of the quadratic function is ; Photovoltaic output The upper and lower limits of It is the energy storage that emits and absorbs power The upper and lower limits of The power consumption of temperature-controlled flexible load The upper and lower limits of It is the temperature setting temperature of the temperature control type flexible load The upper and lower limits of Power consumption of non-temperature-controlled flexible load The upper and lower limits of .

[0255] As a possible implementation of this embodiment, the Lagrange multiplier method is used to solve the optimization model to obtain the final optimization model, including:

[0256] Considering the inequality constraints in equation (26), the Lagrange multiplier method is applied to the optimization model, and the original optimization objective function is transformed into:

[0257] (29),

[0258] In the pair and Lagrange multipliers By taking partial derivatives, we can obtain the optimal conditions for the following equivalent optimization problem:

[0259] (30),

[0260] According to the above optimality equation, we get:

[0261] (31),

[0262] Assuming that all power components in the subnet operate within their inequality constraints, the consistency variables of various power components are defined as:

[0263] (32),

[0264] According to the consistency theory, the consistency variables can only reach consistency after iterative updates. The update formula of the consistency variables of the slave node is:

[0265] (33),

[0266] The update formula of the master node is:

[0267] (34),

[0268] Where M represents the type of regulation resource, is the convergence coefficient, which is closely related to the convergence speed of the consistency variable of the master node; ΔP represents the difference between the total active power demand of all types of loads in the subnet and the active power output of all types in the current subnet;

[0269] In the process of iterative calculation of consistency variables, in order to always satisfy the inequality constraints of various power components, it is necessary to associate the constrained quantity with the consistency variable of the component, so that the constraints are satisfied during the iteration of the consistency variables. Associate the constrained quantity with the consistency variable of the component, and satisfy the constraints during the iteration of the consistency variables:

[0270] (35),

[0271] The temperature constraint and upper and lower limit constraints of power consumption of temperature control flexible load are modified as follows:

[0272] (36),

[0273] The upper and lower power limits of non-temperature-controlled flexible loads are modified to:

[0274] (37),

[0275] The upper and lower limits of photovoltaic output are modified as follows:

[0276] (38),

[0277] The output and storage upper limit constraints of the battery energy storage unit are modified to:

[0278] (39).

[0279] Relying on the current power distribution cloud platform of power grid operators, this paper innovatively proposes a "cloud-partition-end collaborative partitioned and hierarchical new distribution system control architecture" based on the cloud-edge collaborative communication architecture. Figure 6 shown.

[0280] As a possible implementation of this embodiment, step 5, performing hierarchical coordinated control on the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy, includes:

[0281] Based on the cloud-edge collaborative communication architecture, a new partitioned and layered distribution system control architecture with "cloud-partition-end" collaboration is established;

[0282] Based on the new zoned and layered distribution system control architecture, hierarchical coordinated control is carried out by adjusting the objective function according to the source-grid-load-storage self-optimization operation strategy.

[0283] As a possible implementation of this embodiment, Figure 6 As shown, the "end" of the partitioned and layered new distribution system control architecture is responsible for establishing the scheduling model of various distributed flexible resources, as well as data collection and sensing, and communicating real-time operation information with the upper layer, and automatically receiving and responding to the control instructions issued by the "partition" level;

[0284] The "partition" of the partitioned and layered new distribution system control architecture is composed of the control logic layer of each resource cluster, which is responsible for online aggregation of various resources in the cluster, obtaining the cluster aggregation model and uploading it to the cloud management and control platform; at the same time, it responds online to the control instructions issued by the cloud control center and decomposes them into various distributed resources in the group;

[0285] The "cloud" of the new partitioned and layered distribution system control architecture is a cloud control center, which generates cloud resource clusters through online dynamic clustering; collects aggregation models of each cluster and coordinates the regulation capabilities of direct-control resources in different partitions and power grids; and decomposes the dispatching instructions issued by the power grid through cloud command decomposition and sends them to each resource cluster.

[0286] As a possible implementation of this embodiment, the control objective function includes a low network active power loss objective function, a voltage deviation reduction objective function, and an improvement in the photovoltaic power generation consumption objective function in the region.

[0287] Low network active power loss objective function:

[0288] (40),

[0289] Where: N T Indicates the total number of time periods optimized; Represents the collection of all branches in the power distribution system; r ij For branch ij The resistance value; I ij,t For branch ij exist t The current flowing through it at any moment;

[0290] Reduce voltage deviation objective function:

[0291] (41),

[0292] Where: N N Represents the total number of nodes in the power distribution system; U i,t For Node i exist t The voltage amplitude at the moment;

[0293] Improve the objective function of photovoltaic power generation consumption in the region:

[0294] (42),

[0295] In the formula, N PV is the total number of photovoltaic generators in the distribution system; P i,t PV For the i Generator sets in t Actual output at the moment; P i,t PV,max For the i Generator sets in t Maximum output at any time;

[0296] According to the actual needs of distribution network operation optimization, the above optimization objectives are comprehensively considered and the optimization objective function is expressed by weight coefficients:

[0297] (43),

[0298] Where: Represents the weight coefficient, satisfying ; Represents the objective function as shown in equations (40)~(42).

[0299] The constraints of the control objective function include:

[0300] ① Network flow constraints:

[0301] The present invention adopts the Dist-flow method to establish the network flow constraint, and transforms the original model into a convex programming form through second-order cone relaxation to obtain the global optimal solution and good solution speed:

[0302] (44),

[0303] Where: subscripts i and j represent the node numbers in the distribution network radiation network; P gen,j , Q gen,j , P L,j , Q L,j Node j The power supply and load connected at the location have reactive power output; P ij,t and Q ij,t Branch ij The active and reactive power at the head end; v i,t and v j,t Node i, j Voltage; r ij 、x ij Branch ij resistance and reactance; I ij Indicates branch ij the current flowing through it; P jk,t and Q jk,t Branch j - k The active and reactive power at the head end; k :j→k Node j It is the set of child nodes of the parent node; V min 2 and V max 2 They represent the squares of the lower and upper limits of the node voltage operation respectively;

[0304] Output constraints of conventional units:

[0305] (45),

[0306] Tie line transmission power constraints:

[0307] (46),

[0308] in, P Gi,max , P Gi,min The region’s i The upper and lower limits of the active output of the unit, They are the first i The lower upper limit of the transmission power of the tie line and the actual transmission power of the tie line at time t;

[0309] Energy storage constraints:

[0310] (47),

[0311] (48),

[0312] (49),

[0313] in, The state of charge of the energy storage system; is the rated capacity of the energy storage system; is the charging / discharging power of the energy storage system. When the energy storage system discharges to the outside world, , when the outside world charges the energy storage system, , when the energy storage system is neither charging nor discharging, it is in floating charge state. ; They represent the charging efficiency and discharging efficiency of the energy storage system respectively; Respectively represent the minimum and maximum values ​​of the charging / discharging power of the energy storage system; Respectively represent the minimum and maximum value of the state of charge of the energy storage system;

[0314] ⑤ Uncertain photovoltaic output model based on robust cost:

[0315] (50),

[0316] in, They are t Shike Photovoltaic Power Station k The actual output, estimated output, output error, output error limit and uncertainty interval adjustment coefficient; N PV is the total number of photovoltaic stations in the area; G t It is a conservative range measurement of the total uncertainty of photovoltaic output in the region by the global robust cost characterization model. The larger the value, the more conservative the model is.

[0317] like Figure 7As shown, an embodiment of the present invention provides a distributed resource distribution network control device based on dynamic partitioning, including:

[0318] An evaluation system construction module is used to collect distributed resource distribution network data and construct a distribution network partition comprehensive evaluation system, wherein the distribution network partition comprehensive evaluation system includes a source-load matching index, a network loss index, a voltage deviation index, and an electrical distance index;

[0319] The distribution network partitioning module is used to assign weights to indicators based on the combination of hierarchical analysis method and entropy weight method, and to generate distribution network partitioning schemes using the two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation";

[0320] Autonomous output optimization module, used to optimize the autonomous output of distributed power sources based on external consistency control protocol;

[0321] The self-optimizing operation module is used to determine the source-grid-load-storage self-optimizing operation strategy of the distribution network partition based on the distributed consistency algorithm;

[0322] The hierarchical coordination control module is used to perform hierarchical coordination control on the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy.

[0323] The present invention firstly proposes a comprehensive evaluation system for distribution network zoning, which covers 4 categories and 7 indicators for evaluating regional division methods; then, a combined indicator weighting method combining hierarchical analysis method and entropy weight method is proposed, which provides a reasonable and comprehensive weighting method for the weights of regional division indicators; secondly, a distribution network zoning sample generation and screening method based on "Monte Carlo-fuzzy evaluation" two-level sample screening is proposed, which solves the problem that the zoning strategy cannot achieve the global optimal when the sample and data are insufficient; finally, a result-oriented weight optimization and correction method is proposed, which provides diversified zoning strategies that are close to actual needs for different zoning target conditions.

[0324] The present invention proposes a self-optimizing operation system covering "device level-system level". First, a distributed power source grouping method is proposed, and based on the grouping, a device-level self-optimizing operation control method covering external input (following DG) and external input (leading DG) is proposed. Then, a regional source-grid-load-storage self-optimizing operation method is proposed. This method fully considers the regulation cost and resources of distributed resources, and establishes a system-level self-optimizing operation strategy based on the "distributed consistency" theory.

[0325] A computer device provided in an embodiment of the present invention includes a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the device is running, the processor communicates with the memory through the bus, and the processor executes the machine-readable instructions to perform any step of the above-mentioned distributed resource distribution network control method based on dynamic partitioning.

[0326] Specifically, the above-mentioned memory and processor are general-purpose memories and processors. When the processor runs the computer program stored in the memory, it can execute the above-mentioned distributed resource distribution network control method based on dynamic partitioning.

[0327] Corresponding to the method for starting the above-mentioned application, an embodiment of the present invention further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned distributed resource distribution network control methods based on dynamic partitioning are executed.

[0328] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed resource distribution network control method based on dynamic partitioning, characterized in that: The steps include: Step 1: Collect distributed resource distribution network data and build a comprehensive evaluation system for distribution network partitions, where the comprehensive evaluation system includes a source-load matching index, a network loss index, a voltage deviation index, and an electrical distance index; Step 2: weight the indicators based on the combination of analytic hierarchy process and entropy weight method, and use the "Monte Carlo-fuzzy evaluation" two-level sample generation and screening method to generate the distribution network partitioning scheme; Step 3: autonomously optimize the output of distributed power sources based on the external consistency control protocol; Step 4, determining the source-grid-load-storage self-optimization operation strategy of the distribution network partition based on the distributed consistency algorithm; Step 5: Perform hierarchical coordinated control of the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy; The source-load matching index includes the new energy consumption rate of the distribution network when participating in AGC and AVC dispatching: (1), (2), in, D 1 is the new energy consumption rate when the distribution network participates in AGC dispatching, D 2 is the new energy consumption rate when the distribution network participates in AVC dispatch, B is the set of nodes, is the set of distributed renewable energy in node i, and They represent the active output of the flexible resource l at time t when the distribution network responds to AGC and AVC instructions, respectively. represents the maximum active power of flexible resource l at time t; The network loss index includes a first network loss index D 3 and second network loss index D 4: (3), (4), in, r ( i ) is a node i A collection of distributed resources in p i D express i The active power of the node load, p PCC,t express t The active power injected into the grid by the PCC point at the moment; The voltage deviation index includes a first voltage deviation rate index D 5 and the second voltage deviation rate indicator D 6: (5), (6), in, V i,t AGC and V i,t AVC They represent the voltage of node i at time t when the distribution network participates in AGC and AVC regulation, V B Indicates the voltage reference value; The electrical distance indicator includes the average electrical distance d of the partition: (7), in, γ (m) indicates partition m The set of nodes in dd ij Representation Node j The ratio of the voltage change value of the node itself to the voltage change value of node i when a unit reactive power change occurs: (8), in, The voltage-reactive sensitivity matrix is ​​represented by i Line j List.

2. The distributed resource distribution network control method based on dynamic partitioning according to claim 1 is characterized in that: The step 2 is to weight the indicators based on the combination of the hierarchical analysis method and the entropy weight method, and to generate the distribution network partitioning scheme using the two-level sample generation and screening method of "Monte Carlo-Fuzzy Evaluation", including: Based on the combination of analytic hierarchy process and entropy weight method, the weighting and level division of zoning indicators are carried out; The "Monte Carlo-Fuzzy Evaluation" two-level sample generation and screening method is used to generate a set of distribution network partitioning schemes under different working conditions; The distribution network partitioning scheme set is evaluated and optimized to generate the optimal partitioning scheme.

3. The distributed resource distribution network control method based on dynamic partitioning according to claim 2 is characterized in that: The weighting and grading of zoning indicators based on the combination of analytic hierarchy process and entropy weight method include: Defining Metrics jj The subjective weight of j for: (9), in, ii and jj Indicates the index number; I indicates the set of indicators; A jj,ii is the matrix A jj Line ii Column elements, whose meaning is index jj Relative to the indicator ii The importance of A jj,ii The value of is defined as follows: 1 represents the indicator jj Ratio Index ii Important, 0.5 indicates an indicator jj and indicators ii Equally important, 0 means indicator ii Ratio Index jj important; Defining Metrics jj The objective weight of for: (10), in, J Indicates the number of indicators; Indicators jj The information entropy of is calculated as follows: (11), in, N represents the total number of samples participating in the evaluation, Representation sample n For jj The probability of the evaluation level of an indicator is expressed as: (12), in, Representation sample n Medium Indicator jj The evaluation level of The subjective weight and the objective weight are multiplied and combined to finally obtain the index jj The combined weight of for: (13); The indicators are divided into four levels: excellent, good, ordinary and poor, with values ​​of 1, 0.75, 0.5 and 1 respectively. The dividing lines line1~line3 between the four levels are determined by maximizing the numerical differences between the four levels.

4. The distributed resource distribution network control method based on dynamic partitioning according to claim 3 is characterized in that: The two-level sample generation and screening method of "Monte Carlo-Fuzzy Evaluation" is used to generate a set of distribution network partitioning schemes under different working conditions, including: Generate partitioned random samples based on the Monte Carlo method; The fuzzy evaluation method is used to screen the generated samples to obtain partition samples that are more in line with the actual scenario.

5. The distributed resource distribution network control method based on dynamic partitioning according to claim 4 is characterized in that: The step of evaluating and optimizing the distribution network partitioning scheme set to generate the optimal partitioning scheme includes: Preliminary evaluation of samples: First, determine the level dividing lines line1~line3 of each indicator; then generate the indicator-level probability matrix of each sample; then perform combined weighting of indicators; finally, use fuzzy evaluation method to grade and evaluate the samples, and obtain the overall fuzzy evaluation level of each sample; Sample screening and secondary evaluation: First, select excellent samples and re-divide the grade boundaries line1~line3 for excellent samples; then, use the new grade boundaries to re-evaluate the indicators of each excellent sample to obtain a new evaluation level for each sample, and use the entropy weight method to calculate the corresponding objective weight; finally, combine the subjective weight with the new objective weight to obtain the new combined weight when the flexible resource cluster participates in AGC and AVC regulation, and finally obtain the comprehensive evaluation score of each sample participating in AGC and AVC, that is: (19), (20), in, and They represent the comprehensive evaluation scores of sample n corresponding to AGC and AVC, respectively. sample n Medium Indicator jj The new rating level, and Respectively represent the indicators when the flexible resource cluster participates in AGC and AVC regulation jj New portfolio weights; Result-oriented weight optimization and correction: The evaluation results of theoretical calculations are verified for correlation with the actual scheduling effects. If the two show a low correlation, a sequential arrangement matrix of the partition method is constructed based on the actual scheduling effects, and the "Monte Carlo-Fuzzy Evaluation" two-level sample generation and screening method is used to optimize and correct the weights of each indicator.

6. The distributed resource distribution network control method based on dynamic partitioning according to claim 1 is characterized in that: The autonomous output optimization of distributed power sources based on the external consistency control protocol includes: in a distribution network containing multiple DGs, firstly grouping the distributed power sources connected to the distribution network system according to different control targets into voltage control groups and power control groups; then selecting the leading DG and the following DG in the group and the consistency variables, and the leading DG detects the information of the control target of the group; then based on the external consistency protocol, the state consistency of the distributed power sources in the group, the voltage regulation of the distribution network nodes and the power control at the public connection are achieved, so as to realize the self-optimizing operation at the device level.

7. The distributed resource distribution network control method based on dynamic partitioning according to claim 1 is characterized in that: The source-grid-load-storage self-optimization operation strategy for determining distribution network partitions based on a distributed consensus algorithm includes: Establish optimization model and analyze the cost and benefit functions of each part; The Lagrange multiplier method is used to solve the optimization model and obtain the final optimization model.

8. The distributed resource distribution network control method based on dynamic partitioning according to claim 7 is characterized in that: The optimization model establishes an objective function with the best economic performance of the subnet as the goal: (21), In the formula, is the total benefit of the autonomous subnet, They represent the collection of all temperature-controlled flexible loads, non-temperature-controlled flexible loads, new energy units, and battery energy storage units in the autonomous area; , They represent the benefits obtained by the optimal scheduling of the temperature control flexible load and the non-temperature control flexible load participating in the subnet; represents the operating cost of battery energy storage; represents the total cost of photovoltaics; The set temperature for temperature control loads; The power demand of non-temperature-controlled loads; Output active power for photovoltaic; The active power output or stored by the energy storage unit; The cost and benefit functions of each part of the analysis include: Benefit function of temperature control flexible load: (22), In the formula, It is a coefficient used to characterize the benefit of temperature control flexible load. is the outdoor temperature, It is the temperature when the user feels the most comfortable; The electricity benefit function of non-temperature-controlled flexible loads: (23), In the formula, They are k The coefficients of constant, linear and quadratic terms in the benefit function of non-temperature-controlled flexible loads; The operating cost function of photovoltaic itself: (24-1), Photovoltaic abandonment cost function: (24-2), The total cost function of new energy units: (25), In the formula, is the cost coefficient of photovoltaic, wind and solar power abandonment; is the actual output power; is the maximum value of the output power, that is, the output power of the unit when operating in MPPT mode; Battery energy storage cost function: (26), In the formula, For the n The coefficients of the cost function for the battery storage; The operating constraints of each resource are: (27), Power balance constraints in each area: (28), In the formula, Indicates the dispatching instructions issued by the superior dispatching center; It represents the set of common loads in the area except flexible loads; represents the power demand of the oth common load; is the power consumed by the jth temperature-controlled flexible load. There is a quadratic function relationship between the power and the set temperature. The coefficient of the quadratic function is ; Photovoltaic output The upper and lower limits of It is the energy storage that emits and absorbs power The upper and lower limits of The power consumption of temperature-controlled flexible load The upper and lower limits of It is the temperature control type flexible load setting temperature The upper and lower limits of Power consumption of non-temperature-controlled flexible load The upper and lower limits of .

9. The distributed resource distribution network control method based on dynamic partitioning according to claim 8 is characterized in that: The Lagrange multiplier method is used to solve the optimization model to obtain the final optimization model, including: Considering the inequality constraints in equation (26), the Lagrange multiplier method is applied to the optimization model, and the original optimization objective function is transformed into: (29), In the pair and Lagrange multipliers By taking partial derivatives, we can obtain the optimal conditions for the following equivalent optimization problem: (30), According to formula (29), we can get: (31), Assuming that all power components in the subnet operate within their inequality constraints, the consistency variables of various power components are defined as: (32), The update formula of the consistency variable of the slave node is: (33), The update formula of the master node is: (34), Where M represents the type of regulation resource, is the convergence coefficient, which is closely related to the convergence speed of the consistency variable of the master node; ΔP represents the difference between the total active power demand of all types of loads in the subnet and the active power output of all types in the current subnet; Associate the constrained quantity with the consistency variable of the component, and satisfy the constraint during the iteration of the consistency variable: (35), The temperature constraint and upper and lower limit constraints of power consumption of temperature control flexible load are modified as follows: (36), The upper and lower power limits of non-temperature-controlled flexible loads are modified to: (37), The upper and lower limits of photovoltaic output are modified as follows: (38), The output and storage upper limit constraints of the battery energy storage unit are modified to: (39)。 10. The distributed resource distribution network control method based on dynamic partitioning according to any one of claims 1 to 9, characterized in that: The step 5, performing hierarchical coordinated control on the distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy, includes: Establish a new partitioned and layered distribution system control architecture based on the cloud-edge collaborative communication architecture; Based on the new zoned and layered distribution system control architecture, hierarchical coordinated control is carried out by adjusting the objective function according to the source-grid-load-storage self-optimization operation strategy.

11. The distributed resource distribution network control method based on dynamic partitioning according to claim 10 is characterized in that: The "end" of the partitioned and layered new distribution system control architecture is responsible for establishing the dispatching model of various distributed flexible resources, as well as data collection and sensing, and communicating real-time operation information with the upper layer, and automatically receiving and responding to the control instructions issued by the "partition" level; The "partition" of the partitioned and layered new distribution system control architecture is composed of the control logic layer of each resource cluster, which is responsible for online aggregation of various resources in the cluster, obtaining the cluster aggregation model and uploading it to the cloud management and control platform; at the same time, it responds online to the control instructions issued by the cloud control center and decomposes them into various distributed resources in the group; The "cloud" of the new partitioned and layered distribution system control architecture is a cloud control center, which generates cloud resource clusters through online dynamic clustering; collects aggregation models of each cluster and coordinates the regulation capabilities of direct-control resources in different partitions and power grids; and decomposes the dispatching instructions issued by the power grid through cloud command decomposition and sends them to each resource cluster.

12. The distributed resource distribution network control method based on dynamic partitioning according to claim 11 is characterized in that: The control objective function includes a low network active power loss objective function, a voltage deviation reduction objective function, and an improvement in the photovoltaic power generation consumption in the region objective function. Low network active power loss objective function: (40), Where: N T Indicates the total number of time periods optimized; Represents the collection of all branches in the power distribution system; r ij For branch ij The resistance value; I ij,t For branch ij exist t The current flowing through it at any moment; Reduce voltage deviation objective function: (41), Where: N N Represents the total number of nodes in the power distribution system; U i,t For Node i exist t The voltage amplitude at the moment; Improve the objective function of photovoltaic power generation consumption in the region: (42), In the formula, N PV is the total number of photovoltaic generators in the distribution system; P i,t PV For the i Generator sets in t Actual output at the moment; P i,t PV,max For the i Generator sets in t Maximum output at any time; According to the actual needs of distribution network operation optimization, the above optimization objectives are comprehensively considered and the optimization objective function is expressed by weight coefficients: (43), Where: Represents the weight coefficient, satisfying ; Represents the objective function as shown in equations (40)~(42).

13. A distributed resource distribution network control device based on dynamic partitioning, characterized in that: include: An evaluation system construction module is used to collect distributed resource distribution network data and construct a distribution network partition comprehensive evaluation system, wherein the distribution network partition comprehensive evaluation system includes a source-load matching index, a network loss index, a voltage deviation index, and an electrical distance index; The distribution network partitioning module is used to assign weights to indicators based on the combination of hierarchical analysis method and entropy weight method, and to generate distribution network partitioning schemes using the two-level sample generation and screening method of "Monte Carlo-fuzzy evaluation"; Autonomous output optimization module, used to optimize the autonomous output of distributed power sources based on external consistency control protocol; The self-optimizing operation module is used to determine the source-grid-load-storage self-optimizing operation strategy of the distribution network partition based on the distributed consistency algorithm; Hierarchical coordination control module, used to perform hierarchical coordination control on distributed resource distribution network according to the source-grid-load-storage self-optimization operation strategy; The source-load matching index includes the new energy consumption rate of the distribution network when participating in AGC and AVC dispatching: (1), (2), in, D 1 is the new energy consumption rate when the distribution network participates in AGC dispatching, D 2 is the new energy consumption rate when the distribution network participates in AVC dispatch, B is the set of nodes, is the set of distributed renewable energy in node i, and They represent the active output of the flexible resource l at time t when the distribution network responds to AGC and AVC instructions, respectively. represents the maximum active power of flexible resource l at time t; The network loss index includes a first network loss index D 3 and second network loss index D 4: (3), (4), in, r ( i ) is a node i A collection of distributed resources in p i D express i The active power of the node load, p PCC,t express t The active power injected into the grid by the PCC point at the moment; The voltage deviation index includes a first voltage deviation rate index D 5 and the second voltage deviation rate indicator D 6: (5), (6), in, V i,t AGC and V i,t AVC They represent the voltage of node i at time t when the distribution network participates in AGC and AVC regulation, V B Indicates the voltage reference value; The electrical distance indicator includes the average electrical distance d of the partition: (7), in, γ (m) indicates partition m The set of nodes in dd ij Representation Node j The ratio of the voltage change value of the node itself to the voltage change value of node i when a unit reactive power change occurs: (8), in, The voltage-reactive sensitivity matrix is ​​represented by i Line j List.

14. A computer device, characterized in that: It includes a processor, a memory and a bus, the memory stores machine-readable instructions executable by the processor, when the computer device is running, the processor and the memory communicate through the bus, and the processor executes the machine-readable instructions to perform the steps of the distributed resource distribution network control method based on dynamic partitioning as described in any one of claims 1-12.

15. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, executes the steps of the distributed resource distribution network control method based on dynamic partitioning as described in any one of claims 1 to 12.