Power distribution network coordination control method considering distributed power supply and related device

Abstract

The invention discloses a power distribution network coordination control method considering a distributed power supply and a related device, and belongs to a new energy power distribution network regulation and control technology, the method comprises the following steps: establishing a power distribution network state space equation comprising distributed power supply output and control event identification, and reflecting a power grid operation state in real time; generating a dynamic security constraint based on the state vector and the topological structure of the power distribution network by using a pre-trained constraint prediction model; the power distribution network is partitioned, the state vector of each partition is extracted, a partition control strategy is generated according to the partition state and the dynamic security constraint, and local optimal control over an inverter and an energy storage system is achieved; according to the method, all partition control strategies are fused in a global feasible region, a global control strategy is generated, coordination and consistency of partition control are achieved, distributed power supply and load changes can be dynamically responded, the safety, stability and energy scheduling efficiency of a power distribution network are improved, and the method is suitable for a large-scale distributed energy access scene.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of new energy power distribution network regulation, and particularly relates to a power distribution network coordinated control method considering distributed power sources and a related device. BACKGROUND

[0002] With the rapid access of distributed power sources (such as photovoltaic, wind power and other new energy), the operation characteristics of the power distribution network present strong fluctuation, great randomness and high uncertainty. At the same time, as important power electronic devices in the power distribution network, inverters and energy storage systems can not only provide active and reactive regulation capabilities for the power grid, but also realize energy time shift and flexible scheduling of the system through charging and discharging behavior. Therefore, how to effectively coordinate and control the inverters and the energy storage becomes a key problem to ensure the safe, stable and economic operation of the power distribution network.

[0003] At present, the management of the inverters and the energy storage is usually realized through hierarchical control or centralized scheduling methods. For example, one kind of method is based on a local control strategy, and the inverters are independently adjusted according to the voltage / frequency droop characteristics, but this way lacks global coordination and is easy to lead to low resource utilization; another kind of method is to optimize the configuration of the distributed power sources and the energy storage at the global level through a centralized optimization scheduling model, but due to the complex dynamic characteristics of the power distribution network and the strong randomness of the output of the distributed power sources, the traditional optimization method has large calculation amount and insufficient real-time performance, and it is difficult to meet the requirement of fast response. SUMMARY

[0004] Based on this, the application aims to provide a power distribution network coordinated control method considering distributed power sources and a related device, which forms a global optimal control strategy for regulating the inverters and the energy storage system in the power distribution network through state space modeling, event-driven regulation and distributed collaborative optimization.

[0005] In the first aspect, the application provides a power distribution network coordinated control method considering distributed power sources, comprising:

[0006] establishing a state space equation of the power distribution network, wherein the state vector of the state space equation comprises distributed power source output and control event identification;

[0007] generating dynamic security constraints by using a pre-trained constraint prediction model according to the state vector and the topology structure of the power distribution network;

[0008] partitioning the power distribution network, determining the state vector of each power grid partition according to the state vector of the power distribution network, and denoting the state vector of each power grid partition as a partition state vector;

[0009] generating a partition control strategy according to the partition state vector and the dynamic security constraints, and generating a global control strategy according to the partition control strategy and the state vector of the power distribution network, wherein the global control strategy is used to control the inverters and the energy storage system of the power distribution network.

[0010] Further, the establishing the state space equation of the power distribution network comprises:

[0011] obtaining real-time measurement data of the power distribution network, the real-time measurement data comprising node voltages of each node and distributed power outputs;

[0012] generating a control event identifier according to the distributed power outputs and the load by using a pre-trained event trigger model;

[0013] establishing the state space equation of the power distribution network by taking the real-time measurement data and the control event identifier as a state vector.

[0014] Further, the generating the control event identifier according to the distributed power output rate of change and the load by using the pre-trained event trigger model comprises:

[0015] the input of the pre-trained event trigger model is represented as follows:

[0016] ,

[0017] wherein, represents the active power output rate of change of the distributed power, represents the reactive power change rate of the load, represents a time interval;

[0018] the output of the pre-trained event trigger model is represented as:

[0019] ,

[0020] wherein, represents the control event identifier, represents a Sigmoid function, and respectively represent the weight and the hidden state of the event trigger model.

[0021] Further, the generating the dynamic security constraint according to the state vector and the power distribution network topology by using the pre-trained constraint prediction model comprises:

[0022] taking a graph neural network as the network structure of the constraint prediction model, taking the state vector and the power distribution network topology as the input of the constraint prediction model, and outputting an electrical prediction parameter;

[0023] calculating the dynamic security constraint according to the electrical prediction parameter, the dynamic security constraint comprising a voltage constraint and a current constraint.

[0024] Further, the partitioning the power distribution network comprises:

[0025] The spectral clustering algorithm is used to determine the power grid partitions according to the electrical distance matrix and the power grid topology matrix, wherein the electrical distance matrix is used to calculate the Laplacian matrix in the spectral clustering algorithm, and is expressed as follows:

[0026] ,

[0027] represents the Laplacian matrix, represents the degree matrix, represents the electrical distance matrix, represents the impedance of the line Path(i, j), represents the impedance of the point b on the line Path(i, j);

[0028] The power grid partitions obtained by the spectral clustering algorithm are expressed as follows:

[0029] ,

[0030] represents the optimization objective of the spectral clustering characteristic decomposition, represents the matrix trace operation, represents the low-dimensional feature representation obtained by the spectral clustering algorithm characteristic decomposition, and the constraint represents that the vectors are orthogonal to each other and are unitized; represents the characteristic vector threshold, represents the kth power grid partition.

[0031] Further, the partition control strategy is generated according to the partition state vector and the dynamic security constraint, and includes:

[0032] Taking the dynamic security constraint as the constraint of the quadratic programming problem, the partition control strategy is expressed as follows:

[0033] ,

[0034] represents the partition state vector, represents the state reference value, represents the partition control strategy, represents the state deviation weight matrix, represents the control cost coefficient.

[0035] Further, the global control strategy is generated according to the partition control strategy and the state vector of the power grid, and includes:

[0036] The global control objective is constructed according to the partition state vector and the power grid topology structure;

[0037] The global control strategy is obtained by projection gradient fusion according to the partition control strategy and the global control objective.

[0038] In a second aspect, the present invention provides a distribution network coordination control device considering distributed power sources, comprising:

[0039] The distribution network state space establishment module is used to establish the state space equation of the distribution network. The state vector of the state space equation includes the output of distributed generation and the control event identifier.

[0040] The constraint generation module is used to generate dynamic safety constraints based on the state vector and the distribution network topology using a pre-trained constraint prediction model.

[0041] The partition state vector generation module is used to partition the distribution network and determine the state vector of each power grid partition based on the state vector of the distribution network, which is denoted as the partition state vector.

[0042] The control strategy generation module is used to generate partition control strategies based on partition state vectors and dynamic security constraints, and to generate global control strategies based on partition control strategies and distribution network state vectors. The global control strategies are used to control the inverters and energy storage systems of the distribution network.

[0043] Thirdly, the present invention provides an electronic device including a memory storing computer-executable instructions and a processor, which, when executed by the processor, causes the device to perform the steps of the distribution network coordinated control method considering distributed power sources provided in the first aspect.

[0044] Fourthly, the present invention provides a readable storage medium storing a computer-executable program that, when executed, can implement the various steps of the distribution network coordinated control method considering distributed power sources provided in the first aspect.

[0045] The present invention has the following beneficial effects:

[0046] This invention proposes a distribution network coordinated control method and related devices that consider distributed generation. The method takes into account the output fluctuations of distributed generation and the distribution network topology, optimizing the control strategy of the distribution network. By introducing a pre-trained constraint prediction model, dynamic safety constraints can be generated in real time, ensuring the electrical safety of the distribution network under load fluctuations and power supply changes. Partitioning the distribution network effectively reduces the computational complexity of global control, and the fusion optimization of partitioned control strategies and global control strategies improves control efficiency and response speed. Furthermore, combining control event identifiers allows for timely adjustment of the control strategy to cope with system disturbances, thereby enhancing the stability, robustness, and adaptability of the distribution network. The method of this invention achieves efficient and intelligent control of the distribution network, optimizes system performance, reduces energy consumption, and promotes the development of green energy and smart grids. Attached Figure Description

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below only represent some of the embodiments of the present application, and all other drawings obtained by those of ordinary skill in the art without creative effort based on these drawings are within the scope of the present application.

[0048] Figure 1 The implementation flowchart of the power distribution network coordinated control method considering the distributed power supply provided by the embodiment of the present application is shown in the figure.

[0049] Figure 2 The structural schematic diagram of the power distribution network coordinated control device considering the distributed power supply provided by the embodiment of the present application is shown in the figure.

[0050] Figure 3 The electronic device architecture diagram provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only represent some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of the present application.

[0052] Referring to Figure 1 An embodiment of the present application proposes a power distribution network coordinated control method considering the distributed power supply, including the following steps:

[0053] Step S110. Establishing the state space equation of the power distribution network, the state vector of the state space equation including the distributed power supply output and the control event identifier.

[0054] In the power distribution network control, this step first needs to convert the dynamic characteristics of the power distribution network into mathematical equations through the state space model. These equations describe the state evolution of the power distribution network at different time points. The state vector usually includes the distributed power supply output (such as the power generation of solar energy, wind energy, etc.) and the control event identifier. The generation of the control event identifier is closely related to the fluctuations of the distributed power supply and the changes of the load. By obtaining real-time measurement data, combining historical data and event triggering models, the state space equation is constructed to ensure that the state of the system can accurately reflect the current operation of the power grid.

[0055] The control event identifier introduced in this step refers to the control event determined by the event triggering model during the operation of the power distribution network, which is used to identify whether the system needs to adjust the control strategy. This identifier helps to determine when to issue control instructions, avoiding too frequent control adjustments.

[0056] Specifically, in the distribution network, the state vector usually includes information such as voltage, current, power, load and distributed power output, etc., and the distributed power output refers to the power output of the distributed power generation system (such as solar energy, wind energy, micro gas power generation, etc.), which is an important variable in the operation of the distribution network.

[0057] Further, the state vector includes line power (active and reactive), which is directly obtained in the preferred embodiment if the line installation measurement device is installed, otherwise it is calculated by state estimation. For example, the state estimation calculation of the line active power and the line reactive power is given as follows:

[0058]

[0059] wherein, and represent the line active power and the line reactive power, respectively, and represent the node voltage, and represent the line admittance, i=j represents the self-impedance or self-admittance, and i≠j represents the mutual impedance or mutual admittance, represent the node phase angle difference.

[0060] Further, the control event identification is generated according to the distributed power output rate of change and the load by using the pre-trained event trigger model.

[0061] In the traditional distribution network control, fixed cycle sampling and periodic control method are usually used, which may lead to frequent triggering of control, even if the system is in a stable state, additional communication and calculation resources will be consumed, and the control action will be delayed, which cannot quickly respond to the sudden fluctuations of distributed power and load.

[0062] To solve these problems, the pre-trained event trigger model is introduced in the present scheme, which usually uses a lightweight neural network structure (such as RNN, GRU or LSTM) to capture the time sequence characteristics, or uses a fully connected neural network (MLP) for feature mapping.

[0063] Specifically, the input of the pre-trained event trigger model is represented as follows:

[0064]

[0065] wherein, represents the active power output rate of change of the distributed power, represents the reactive power change rate of the load, represents the time interval.

[0066] The model hidden layer is activated by a single-layer tanh function, and the weights and biases can be determined by training historical data. The final output is between 0 and 1, representing the probability or intensity of triggering control under the current state.

[0067] The output of the pre-trained event trigger model can be represented as:

[0068]

[0069] where, represents the control event identifier, represents the Sigmoid function, and represent the weights and hidden states of the event trigger model, respectively.

[0070] The model training data can use historical event libraries, such as photovoltaic sudden drop, load surge, etc. corresponding control event identifier true value. By pre-training the event trigger model, random fluctuations (such as photovoltaic cloud disturbance) are encoded as event triggers, replacing fixed time window sampling, similar to the interrupt wake-up mechanism in embedded programming, which can effectively capture dynamic events and effectively reduce hardware consumption and power consumption.

[0071] The introduction of the control event identifier makes the state vector not only represent the physical state of the system, but also carry the logical information of the "control opportunity", so that the state space equation can reflect both "electrical dynamics" and "control dynamics", and is more close to the actual operation; secondly, with the event identifier, control will only intervene when the trigger condition is met, avoiding frequent switching of inverters and energy storage, prolonging the service life of equipment; thirdly, when the distributed power output or load appears sudden and large fluctuations, the event identifier can quickly trigger control to ensure grid stability, and can also consider the active or reactive power output changes of distributed power sources, avoiding single signal misjudgment.

[0072] Step S120. Use the pre-trained constraint prediction model to generate dynamic security constraints based on the state vector and the distribution network topology.

[0073] Dynamic security constraints refer to real-time updated constraints in the operation of distribution networks, usually including maximum or minimum limits of voltage, power, frequency, etc. These constraints ensure the stability and safety of distribution networks under various load changes and power fluctuations.

[0074] The step utilizes a constraint prediction model combined with the state vector and topology structure of the power distribution network. The system can automatically generate dynamic safety constraints, which help to control the power distribution network safely and prevent problems such as overload, short circuit, voltage drop, etc. For example, the upper and lower limits of voltage and current are important dynamic safety constraints that must be strictly followed in power distribution network control. The constraint prediction model is based on past operation data, weather conditions, load forecasts, and other information to predict potential risks of the power grid in the current state and generate corresponding safety limits.

[0075] Further, the graph neural network is used as the network structure of the constraint prediction model, and the state vector and the topology structure of the power distribution network are used as the input of the constraint prediction model. The electrical prediction parameters are output. The dynamic safety constraints are calculated according to the electrical prediction parameters, and the dynamic safety constraints include voltage constraints and current constraints.

[0076] The graph neural network (GNN) is used to jointly encode the topology relationship (nodes and branches) of the power distribution network and the time-varying state vector of each node / branch. The local and global electrical coupling relationship is learned through message passing, and the electrical prediction parameters used to calculate the dynamic safety constraints (such as node voltage upper and lower limits and line current limits) are output, thereby realizing the generation of safety constraints based on real-time state adaptation.

[0077] Specifically, the prediction model is represented as follows:

[0078]

[0079] wherein, represents the state vector, which is used to provide the real-time state; represents the topology correlation matrix, which is used to define the topology structure of the power distribution network, and the matrix elements represent that node i is directly connected to node j, otherwise 0; GNN aggregates neighborhood information through graph convolution and outputs dynamic constraints .

[0080] Compared with the fixed value constraints used in traditional methods, the prediction model can adaptively adjust with fluctuations in distributed power sources (such as relaxing the voltage upper limit when photovoltaic output is high).

[0081] The output of the GNN is represented as:

[0082]

[0083] represents the predicted mean value of node voltage, represents the predicted standard deviation of node voltage, represents the gradient descent step size, and these electrical prediction parameters output by the GNN are used to calculate the voltage constraints and current constraints.

[0084] The linear output can be directly reflected by the voltage expectation value; The Softplus activation function is used to ensure positive values, which can be expressed as:

[0085]

[0086] where, The neighborhood node information is aggregated by graph convolution for each layer, and

[0087]

[0088] The degree matrix is represented as, The trainable weight matrix of the convolution layer l is represented as.

[0089] The dynamic security constraint can be calculated as follows:

[0090]

[0091] and The lower voltage limit and the upper voltage limit are represented as, The line current constraint is represented as, The rated current is represented as, The 95% confidence interval is covered, Adjusted by GNN.

[0092] Further, the constraint prediction time limit model loss function is fused with the physical equation as follows:

[0093]

[0094] Physical consistency:

[0095]

[0096]

[0097] Convexity guarantee:

[0098]

[0099]

[0100] where is the Jacobian matrix of the power flow equation, P DG The active power output of the distributed power supply is represented as, is the weight coefficient, is a small constant that can be set to 0.01, is a small perturbation, and the typical value is 0.01 p.u.

[0101] By matching the power flow item Forcing GNN to output satisfying power flow equation (i.e. power balance), ensuring the physical reasonableness of prediction constraints; curvature regularization term Ensuring convexity, ensuring constraints Decoupling voltage / current constraints in and power flow equation, avoiding the prediction result from deviating from the actual power grid operation state, ensuring that the curvature term guarantees the constraints The feasible region defined is a convex set, which is mathematically solvable.

[0102] Step S130. Partition the distribution network, determine the state vector of each power grid partition according to the state vector of the distribution network, denoted as the partition state vector.

[0103] The purpose of partitioning the distribution network is to divide the distribution network into multiple relatively independent regions, so as to effectively control in a local range. The partition is based on the topology of the distribution network, and the electrical distance matrix and the topology matrix are analyzed by using methods such as spectral clustering algorithm, to ensure that the control of each region does not interfere with each other and conforms to the operation characteristics of the power grid. The state vector of each partition represents the electrical parameters in the region and can reflect the operation state of the partition. Reasonable partitioning makes the generation and execution of control strategies more efficient, reducing the complexity of global optimization.

[0104] Specifically, the spectral clustering algorithm is used to determine the power grid partitions according to the electrical distance matrix and the distribution network topology matrix, wherein the electrical distance matrix is used to calculate the Laplacian matrix in the spectral clustering algorithm, which is represented as follows:

[0105]

[0106] represents the Laplacian matrix, represents the degree matrix, represents the electrical distance matrix, represents the impedance of line Path(i, j), represents the impedance of point b on line Path(i, j);

[0107] The spectral clustering algorithm is decomposed to obtain the power grid partitions, which are represented as follows:

[0108]

[0109] represents the optimization objective of spectral clustering feature decomposition, represents the matrix trace operation, represents the low-dimensional feature representation obtained by spectral clustering algorithm feature decomposition, constraint represents that the two vectors are orthogonal and unitized; represents the feature vector threshold, represents the kth power grid partition.

[0110] Step S140. Generate a partitioned control strategy based on the partitioned state vector and dynamic security constraints, and generate a global control strategy based on the partitioned control strategy and the state vector of the distribution network. The global control strategy is used to control the inverters and energy storage systems of the distribution network.

[0111] Based on the state vector and dynamic security constraints of each partition, this step generates a control strategy for each partition. Using quadratic programming, the system calculates the optimal control command for each partition's state vector and constraints. These partition control strategies are then further integrated to generate a global control strategy based on the global objective. This global control strategy coordinates the behavior between partitions, optimizing the overall operation of the distribution network.

[0112] Specifically, within each grid partition, based on the partition state vector (including node voltage, distributed generation output, energy storage SOC, control event identifiers, etc.) and the dynamic safety constraint set given in step S120 (node ​​voltage upper and lower limits, branch current upper limits, etc.), a partition control strategy is generated for the controllable devices (inverters, energy storage) within the partition. The strategy generation method uses quadratic programming (QP) as the basic optimization kernel, aiming to achieve the following under the premise of ensuring safety constraints: state tracking / steady-state recovery, loss or cost minimization, inverter and energy storage operation smoothness constraints, and meeting event triggering requirements.

[0113] Using dynamic safety constraints as the constraints of the quadratic programming problem, the zoning control strategy can be expressed as the following quadratic programming problem:

[0114]

[0115] Represents the partition state vector. Indicates the state reference value. Indicates the partition control strategy. Represents the state deviation weight matrix. This represents the control cost coefficient.

[0116] The constraints of the planning problem are expressed as follows:

[0117]

[0118] The neural constraint projection is a mathematical operation that forces the control quantity to conform to the feasible region predicted by the GNN, and forces the optimization solution to be mapped to the feasible region predicted by the neural network. Its projection expression is as follows:

[0119]

[0120] in The original solution of the quadratic programming output, which may violate dynamic constraints, is modified by projection to ensure the control instruction is safe and feasible.

[0121] The state tracking term in the planning problem forces the partition state (e.g. voltage) to approach the reference value (usually the rated value); The control cost term is used to minimize the amplitude of device action (e.g. reduce the frequent charging and discharging of energy storage).

[0122] The global control objective is constructed according to the partition state vector and the power grid topology as follows:

[0123]

[0124] The loop current suppression term penalizes the power imbalance at the sub-domain boundary (e.g. the adjacent inverter adjusts in the opposite direction) through the topological incidence matrix T, and K represents the total number of grid partitions.

[0125] The projection gradient fusion algorithm takes the partition control strategy as the initial value, performs gradient fusion / correction on the global feasible region, and ensures safety and feasibility by projecting onto the global feasible set or local feasible set at each step.

[0126] The basic idea of the projection gradient fusion algorithm is to start from the summation of the partition control strategies , and iterate along the negative gradient direction of the objective function. After each step, the result is projected back to the feasible set to obtain a solution that gradually converges to a solution that takes into account both the partition recommendations and global constraints.

[0127] The global control strategy is represented as follows:

[0128]

[0129] The gradient step size of the projection gradient fusion process is represented by , which pushes the global solution to converge to the optimal objective (e.g. voltage stability, minimum network loss), and the projection operation forces the result to satisfy the dynamic safety constraints (avoid over-limit), The global feasible region is represented as:

[0130] F={u* / Φ GNN (x k ,T)≥0}

[0131] The significance lies in converting the aforementioned dynamic safety constraints into a mathematical feasible region, ensuring that the global control strategy satisfies the voltage constraint, current constraint, and device capacity constraint at the same time.

[0132] Final global control strategy The global control strategy can be used to control the reactive power output of the inverters to achieve voltage support, and control the energy storage system to perform active smoothing power fluctuations, and further can be used to guide flexible loads to participate in demand side management.

[0133] Further, the global control strategy can be used to select a specific device for adjustment, for example, selecting a device with the highest voltage / power regulation sensitivity, or selecting a device with the lowest device regulation cost, while excluding devices in fault / maintenance, and in a preferred embodiment, the selection of the device can be determined by the following score:

[0134]

[0135] denotes the score of the i-th device, , , denote the sensitivity, economy and health, respectively, , , denote the weight coefficients.

[0136] The above-mentioned method can be implemented in various forms of devices, and therefore the present application also discloses an apparatus corresponding to the above-mentioned method, and the following specific embodiments are given to explain in detail.

[0137] As shown in Figure 2 , one embodiment of the present application provides a power distribution network coordinated control device considering distributed power supply, comprising:

[0138] A power distribution network state space establishment module 202 is configured to establish a state space equation of the power distribution network, and a state vector of the state space equation comprises distributed power supply output and control event identification;

[0139] A constraint generation module 204 is configured to generate dynamic security constraints according to the state vector and the power distribution network topology by using a pre-trained constraint prediction model;

[0140] A partition state vector generation module 206 is configured to partition the power distribution network, and determine a state vector of each power grid partition according to the state vector of the power distribution network, denoted as a partition state vector;

[0141] A control strategy generation module 208 is configured to generate a partition control strategy according to the partition state vector and the dynamic security constraints, and generate a global control strategy according to the partition control strategy and the state vector of the power distribution network, wherein the global control strategy is used to control the inverters and the energy storage system of the power distribution network.

[0142] ​The device provided by the embodiments of the present application has the same implementation principle and generated technical effects as the foregoing method embodiments. For brevity of description, the part of the device embodiments not mentioned can be referred to the corresponding content in the foregoing method embodiments.

[0143] The method and related device mentioned in each of the foregoing embodiments are described with reference to the method flowchart and / or structural schematic diagram provided by the embodiments of the present application. Each flow and / or block in the method flowchart and / or structural schematic diagram and the combination of the flows and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. The computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate a method implemented in the flowchart and / or block diagram. Figure 1 The computer program instructions can also be stored in a computer-readable memory capable of causing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a product including instruction devices, which implement the functions specified in the flowchart and / or block diagram. Figure 1 The computer program instructions can also be stored in a computer-readable memory capable of causing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a product including instruction devices, which implement the functions specified in the flowchart and / or block diagram. Figure 1 The computer program instructions can also be stored in a computer-readable memory capable of causing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a product including instruction devices, which implement the functions specified in the flowchart and / or block diagram. Figure 1 The computer program instructions can also be stored in a computer-readable memory capable of causing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a product including instruction devices, which implement the functions specified in the flowchart and / or block diagram. Figure 3 The computer program instructions can also be stored in a computer-readable memory capable of causing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a product including instruction devices, which implement the functions specified in the flowchart and / or block diagram.

[0144] The following embodiments take the computer device as an example, which can be any device with operation and processing functions, such as a server or a personal notebook computer. In one of the embodiments, the computer device can be an application server, which can be a server for running an application to be tested.

[0145] For brevity of description, the part of the device embodiments not mentioned can be referred to the corresponding content in the foregoing method embodiments. Figure 3FIG. 1 illustrates a hardware structure block diagram of an electronic device, which is intended to represent various forms of digital computers, such as laptops, desktops, tablets, servers, servers, servers, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0146] As shown in ​ The electronic device includes at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4;

[0147] In the embodiments of the present application, the number of processors 1, communication interfaces 2, memories 3, and communication buses 4 is at least one, and the processors 1, communication interfaces 2, and memories 3 communicate with each other through the communication bus 4;

[0148] The processor 1 can be a central processing unit CPU, or an application specific integrated circuit ASIC, or one or more integrated circuits configured to implement the embodiments of the present application, etc.

[0149] The memory 3 can include a high-speed RAM memory, and can also include a non-volatile memory, etc., such as at least one disk memory.

[0150] The memory stores a program, and the processor can call the program stored in the memory, and the program is used to implement the various processing procedures of the power distribution network coordination control scheme considering distributed power sources.

[0151] The embodiments of the present application also provide a readable storage medium having a computer program stored thereon, and the computer program is executed by the processor to implement the various processing procedures of the power distribution network coordination control scheme considering distributed power sources provided by the above embodiments and / or any one of the possible implementation manners combined with the embodiments.

[0152] The above-described embodiments of the application have been described in connection with what are presently considered to be the most practical and preferred implementations, it will be apparent to those of ordinary skill in the art that numerous modifications, implementations, and equivalents can be made without parting from the spirit and scope of the application. The specific naming of the components, capitalization of terms, the attributes, data structures, or any other programming or structural aspect is not mandatory or significant, and the mechanisms that implement the application or its features can have different names, formats, or protocols. The illustrated embodiments are described in enough detail to enable those with ordinary skill in the art to practice the application. It will be apparent to those of ordinary skill in the art that numerous implementations can be made without departing from the scope of the application. It is intended to include all such modifications, enhancements, alternatives, permutations, and equivalents as can be included within the spirit and scope of the application. The following claims are in no way intended to limit the scope of the present application to the precise

[0153] Those skilled in the art will appreciate that the various steps of the methods disclosed above can be implemented by general purpose computing devices, which can be centralized on a single computing device or distributed across a network of multiple computing devices, and optionally implemented in program code executable by a computing device, which can be stored in a storage device and executed by a computing device, or alternatively implemented as individual integrated circuit modules, or multiple modules or steps implemented as a single integrated circuit module. Thus, the embodiments of the present application are not limited to any particular combination of hardware and software.

[0154] The computing device executable programs (also referred to as programs, software, software applications, or code) include machine instructions for a programmable processor, and can be implemented in a high-level procedural and / or object-oriented programming language, and / or in assembly / machine language. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus and / or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0155] Certain aspects of the application include the processes described herein, and instructions to carry out those processes, in the form of algorithm. It is to be understood that the processes and instructions of the application can be embodied in software, firmware and / or hardware, and when implemented in software, can be downloaded to reside on and be operated from different platforms used by a variety of operating systems.

[0156] Those skilled in the art can understand that the structure shown in each figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the terminal device to which the scheme of the present application is applied. The specific terminal device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0157] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "a possible design" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine the different embodiments or examples described in the present specification and the features of the different embodiments or examples without contradiction.

[0158] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distribution network coordinated control method considering distributed generation, characterized in that, include: Establish the state-space equations of the distribution network, wherein the state vectors of the state-space equations include distributed generation output and control event identifiers; Dynamic security constraints are generated based on the state vector and the distribution network topology using a pre-trained constraint prediction model. The distribution network is divided into zones, and the state vector of each power grid zone is determined based on the state vector of the distribution network, denoted as the zone state vector; A partition control strategy is generated based on the partition state vector and dynamic security constraints, and a global control strategy is generated based on the partition control strategy and the distribution network state vector. The global control strategy is used to control the inverters and energy storage systems of the distribution network.

2. The method according to claim 1, characterized in that, The state-space equations for establishing the power distribution network include: Acquire real-time measurement data of the distribution network, including node voltage and distributed power output of each node; A pre-trained event-triggered model is used to generate control event identifiers based on the output change rate of distributed power sources and the load. The state-space equations of the distribution network are established using real-time measurement data and control event identifiers as state vectors.

3. The method according to claim 2, characterized in that, The pre-trained event-triggered model utilizes distributed power generation output change rate and load generation control event identifiers, including: The input representation of the pre-trained event-triggered model is as follows: ， in, This represents the rate of change in the active power output of distributed generation. Indicates the rate of change of reactive power of the load. Indicates a time interval; The output of the pre-trained event-triggered model is represented as: ， in, Indicates the control event identifier, This represents the Sigmoid function. and These represent the weights and hidden states of the event-triggered model, respectively.

4. The method according to claim 1, characterized in that, The generation of dynamic security constraints using a pre-trained constraint prediction model based on the state vector and the distribution network topology includes: A graph neural network is used as the network structure of the constrained prediction model, and the state vector and distribution network topology are used as the inputs of the constrained prediction model. The output is the electrical prediction parameters. Dynamic safety constraints are calculated based on the electrical prediction parameters. These dynamic safety constraints include voltage constraints and current constraints.

5. The method according to claim 1, characterized in that, The zoning of the distribution network includes: The spectral clustering algorithm is used to determine each power grid zone based on the electrical distance matrix and the distribution network topology matrix. The electrical distance matrix is ​​used to calculate the Laplace matrix in the spectral clustering algorithm, as shown below: ， Represents the Laplace matrix, Degree matrix, Represents the electrical distance matrix. This represents the impedance of the line path(i, j). This represents the impedance of point b on line Path(i, j); The eigenvalue decomposition obtained by the spectral clustering algorithm yields the following power grid partition representation: ， This represents the optimization objective of spectral clustering eigenvalue decomposition. Represents the matrix trace operation. This represents the low-dimensional feature representation obtained by eigenvalue decomposition in the spectral clustering algorithm, with constraints. This indicates that the vectors are pairwise orthogonal and normalized. Represents the threshold of the feature vector. This represents the k-th power grid partition.

6. The method according to claim 1, characterized in that, The step of generating a partition control strategy based on the partition state vector and dynamic security constraints includes: Using dynamic safety constraints as the constraints of the quadratic programming problem, the zoning control strategy can be expressed as the following quadratic programming problem: ， Represents the partition state vector. Indicates the state reference value. Indicates the partition control strategy. Represents the state deviation weight matrix. This represents the control cost coefficient.

7. The method according to claim 1, characterized in that, The step of generating a global control strategy based on the zonal control strategy and the state vector of the distribution network includes: Construct a global control objective based on the state vectors of each zone and the distribution network topology; The global control strategy is obtained by performing projection gradient fusion based on the partition control strategy and the global control objective.

8. A distribution network coordination control device considering distributed generation, characterized in that, include: The distribution network state space establishment module is used to establish the state space equation of the distribution network. The state vector of the state space equation includes distributed generation output and control event identifiers. The constraint generation module is used to generate dynamic safety constraints based on the state vector and the distribution network topology using a pre-trained constraint prediction model. The partition state vector generation module is used to partition the distribution network and determine the state vector of each power grid partition based on the state vector of the distribution network, which is denoted as the partition state vector. The control strategy generation module is used to generate a partition control strategy based on the partition state vector and dynamic security constraints, and to generate a global control strategy based on the partition control strategy and the state vector of the distribution network. The global control strategy is used to control the inverters and energy storage systems of the distribution network.

9. An electronic device, characterized in that, It includes a memory storing computer-executable instructions and a processor, which, when executed by the processor, causes the device to perform the distribution network coordinated control method considering distributed power sources as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that, It stores a computer-executable program that, when executed, enables the distribution network coordinated control method considering distributed generation as described in any one of claims 1 to 7.

Images