Power distribution network regulation and control method and system

Abstract

The invention discloses a power distribution network regulation and control method and system. The method comprises the steps that new energy output data, obtained through prediction, of a target power distribution network are acquired; inputting the new energy output data into the power distribution network operation model to obtain power flow data of each node; sensitivity analysis is carried out on the power flow data, clustering is carried out based on a sensitivity analysis result, and nodes are divided into a plurality of initial node groups; correcting the initial node group based on the topological structure information to obtain a plurality of target node groups; respectively inputting the power flow data into a pre-constructed mobile emergency power supply scheduling model and a hybrid intelligent soft switch regulation and control model, and generating a primary regulation and control scheme according to an output result; calculating a reactive power demand of the target node group according to the power flow data and the real-time voltage data, and generating a secondary regulation and control scheme according to the reactive power demand; and executing the primary regulation and control scheme and the secondary regulation and control scheme. According to the method provided by the invention, the real-time voltage regulation and control requirement of the power distribution network is met.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network optimization, and in particular to a power distribution network control method and system. Background Art

[0002] In new power systems, reactive power regulation is crucial to grid reliability and voltage stability. With the large-scale access of distributed renewable energy, the frequency and rapidity of short-term voltage limit violations in distribution networks have become increasingly prominent, which makes it difficult for traditional reactive power-voltage optimization methods to meet the real-time voltage regulation needs of new distribution networks.

[0003] Although traditional reactive power compensation devices, such as static VAR compensators and traditional capacitors, have been widely used in distribution networks, they are often unable to fully adapt to the rapid and frequent changes in distribution network voltage, resulting in the voltage stability of the distribution network being threatened when encountering load disturbances.

[0004] It can be seen that how to optimize the control method of the distribution network has become a technical problem that needs to be solved urgently by technical personnel in this field. Summary of the invention

[0005] The present invention provides a distribution network control method and system to solve the technical problem of how to optimize the control of the distribution network.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides a distribution network control method, including: Acquire predicted new energy output data of the target distribution network, input the new energy output data into a pre-built distribution network operation model, and obtain power flow data of each node in the target distribution network; Performing sensitivity analysis on the power flow data, performing clustering based on the sensitivity analysis result, and dividing the nodes into a plurality of initial node groups, wherein the clustering is designed to perform node division according to the influence of reactive power on node voltage and the electrical distance between nodes; Based on the acquired topological structure information of the target distribution network, the initial node group is modified to obtain a plurality of target node groups matching the actual physical structure; The power flow data are respectively input into a pre-built mobile emergency power supply dispatching model and a hybrid intelligent soft switch control model, and a primary control scheme of the target distribution network is generated according to the output results; Calculating the reactive power demand of the target node group according to the power flow data and the real-time voltage data of the target distribution network, and generating a secondary control scheme of the target distribution network according to the reactive power demand; Execute the primary control scheme and the secondary control scheme.

[0007] As one of the preferred solutions, the respectively obtaining predicted new energy output data of the distribution network includes: Construct an initial output prediction model based on deep neural network; The initial output prediction model is trained based on the acquired historical renewable energy output data and historical weather data to obtain a trained target output prediction model; In the process of predicting the new energy output of the target distribution network, real-time weather data is input into the target output prediction model to obtain the new energy output data.

[0008] As one of the preferred solutions, the training of the initial output prediction model based on the acquired historical renewable energy output data and historical weather data further includes: During the training process, the hyperparameters for constructing the initial output prediction model are optimized according to the genetic algorithm, specifically including: Randomly generate a set of hyperparameter combinations as an initial population, where each individual in the initial population represents a set of different hyperparameter values; The individual is evaluated using the initial output prediction model, and the fitness value of the individual is calculated according to the pre-constructed fitness function. The fitness value is expressed as: in, is the fitness value, , , is the weight factor; The selection, crossover and mutation operations are iteratively performed according to the fitness value until a preset number of iterations is reached to obtain the optimal individual of the current population, so as to complete the optimization of the hyperparameters of the initial output prediction model.

[0009] As one of the preferred solutions, the step of inputting the new energy output data into a pre-built distribution network operation model to obtain the flow data of each node in the target distribution network includes: Establishing an initial distribution network operation model based on the acquired load parameters of the target distribution network; Optimizing the parameters of the target distribution network operation model according to a time series analysis method to obtain the target distribution network operation model; The new energy output data is input into the target distribution network operation model to obtain the flow data of each node in the target distribution network.

[0010] As one of the preferred solutions, the sensitivity analysis result includes a sensitivity joint index, which is expressed as: in, , are node phase angle and voltage change vector respectively, , are the node active and reactive power change vectors, is the reactive power sensitivity matrix of the node, and Indicates the voltage change value corresponding to grid nodes i and j at time t; The correlation coefficient of the voltage change at node j causing the voltage change at node i is, The correlation coefficient indicating the voltage change at node i causing the voltage change at node j; represents the electrical distance between nodes i and j; is the reactive power sensitivity; is the sensitivity joint index, is the weight.

[0011] As one of the preferred solutions, the initial node group is modified based on the acquired topological structure information of the target distribution network to obtain multiple target node groups matching the actual physical structure, including: Acquire topological structure information of the target distribution network, wherein the topological structure information includes node data, line connection relationship data and electrical area division data; The nodes in the initial node group are adjusted according to the topological structure information to obtain a plurality of target node groups matching the actual physical structure.

[0012] As one of the preferred solutions, clustering is performed based on the sensitivity analysis result to divide the nodes into multiple initial node groups, including: Each node is regarded as an independent cluster, and the sensitivity joint index dataset is constructed based on the sensitivity joint index of all clusters; In the node merging process, two clusters with the smallest distance between the sensitivity joint indicator data sets are merged to obtain the current sensitivity joint indicator data set; The node merging process is iterated until a preset number of clusters is met, so as to divide the nodes into a plurality of initial node groups.

[0013] As one of the preferred solutions, the output results include grid load, photovoltaic power generation forecast data, real-time voltage fluctuation, operating cost and loss cost; The power flow data is respectively input into the pre-built mobile emergency power dispatch model and the hybrid intelligent soft switch control model, and a primary control scheme of the target distribution network is generated according to the output results, including: Dynamically adjusting the location and power distribution of the mobile emergency power supply according to the grid load, the photovoltaic power generation forecast data and the real-time voltage fluctuation; And dynamically adjust the transmission power of the intelligent soft switch and the tie switch according to the operating cost and loss cost.

[0014] As one of the preferred solutions, the secondary control scheme for generating the target distribution network according to the reactive power demand includes: Establishing a control target according to the reactive power demand, and formulating a control plan corresponding to the control target; The control scheme is iteratively optimized based on a particle swarm optimization algorithm to obtain a secondary control scheme for the target distribution network.

[0015] Another embodiment of the present invention provides a distribution network control system, including: An acquisition module is used to acquire predicted new energy output data of a target distribution network, input the new energy output data into a pre-built distribution network operation model, and obtain power flow data of each node in the target distribution network; An analysis module, configured to perform sensitivity analysis on the power flow data, perform clustering based on the sensitivity analysis result, and divide the nodes into a plurality of initial node groups, wherein the clustering is designed to perform node division based on the influence of reactive power on node voltage and the electrical distance between nodes; A correction module, used to correct the initial node group based on the acquired topological structure information of the target distribution network to obtain multiple target node groups matching the actual physical structure; A first control module is used to input the power flow data into a pre-built mobile emergency power supply scheduling model and a hybrid intelligent soft switch control model, and generate a primary control scheme of the target distribution network according to the output results; A second control module is used to calculate the reactive power demand of the target node group according to the power flow data and the real-time voltage data of the target distribution network, and generate a secondary control scheme of the target distribution network according to the reactive power demand; An execution module is used to execute the primary control scheme and the secondary control scheme.

[0016] Compared with the prior art, the embodiments of the present invention have the following advantages: 1) The present invention obtains the predicted target distribution network new energy output data and inputs it into the distribution network operation model to obtain the flow data of each node, so as to more accurately grasp the operation status of the distribution network. The uncertainty of new energy output is an important factor affecting the operation of the distribution network. Obtaining and considering its output data in advance can make the flow calculation more in line with the actual operation situation, provide an accurate data basis for subsequent regulation, and thus improve the overall efficiency and reliability of the distribution network operation.

[0017] 2) This invention performs sensitivity analysis on power flow data and clusters and divides node groups according to the impact of reactive power on node voltage and the electrical distance between nodes, which helps to gain a deeper understanding of the distribution and transmission characteristics of reactive power in the distribution network. The reactive power control scheme formulated based on this is more scientific and reasonable, which can effectively reduce the unreasonable flow of reactive power, reduce network losses, improve voltage stability, and further enhance the reliability and economy of distribution network operation.

[0018] 3) The present invention introduces a mobile emergency power supply scheduling model and a hybrid intelligent soft switch control model to generate a primary control plan, and generates a secondary control plan according to reactive power demand, providing a multi-level and multi-means control strategy for the distribution network. The mobile emergency power supply and hybrid intelligent soft switch are flexible and responsive, and can quickly adjust the power flow distribution and voltage level of the distribution network under different operating conditions, such as sudden changes in new energy output and load peak and valley changes, effectively alleviating the impact of new energy access on the distribution network and ensuring the stable operation of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flow chart of a distribution network control method in one embodiment of the present invention; Figure 2 is a flowchart of deep neural network model hyperparameter optimization in one embodiment of the present invention; Figure 3 is a flow chart of distribution network hierarchical processing in one embodiment of the present invention; Figure 4 is a schematic diagram of the structure of a hybrid intelligent soft switch in one embodiment of the present invention; Figure 5 It is a schematic diagram of a distribution network control system in one embodiment of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0022] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used herein are only for illustrative purposes, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0023] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0024] An embodiment of the present invention provides a distribution network control method. For details, see Figure 1 , Figure 1 The figure shows a flow chart of a distribution network control method in one embodiment of the present invention, which includes steps S1-S6: S1: Obtain the predicted new energy output data of the target distribution network, input the new energy output data into the pre-built distribution network operation model, and obtain the flow data of each node in the target distribution network; In this step, a deep neural network model based on genetic algorithm optimization of hyperparameters is used to analyze and predict the changes in photovoltaic power generation and load in the distribution network by constructing a mapping relationship between season, time period, photovoltaic power generation / load historical data and photovoltaic power generation / load forecast data.

[0025] Specifically, collect relevant historical data, including seasonal changes, time of day and other characteristic variables that affect photovoltaic power generation / load demand and historical photovoltaic power generation / load data. Perform preprocessing such as cleaning, missing value filling, outlier processing, and normalization on these historical data to ensure the quality of the model's input data. Determine the required prediction time scale (which can be minute or hour level), and then build a deep neural network model based on the programming language to obtain the mapping association between the photovoltaic power generation / load data before the season, time period, and time scale and the photovoltaic power generation / load data after the time scale.

[0026] Preferably, in one embodiment of the present invention, respectively obtaining the predicted new energy output data of the distribution network includes: Construct an initial output prediction model based on deep neural network; The initial output prediction model is trained based on the acquired historical renewable energy output data and historical weather data to obtain a trained target output prediction model; In the process of predicting the renewable energy output of the target distribution network, the real-time weather data is input into the target output prediction model to obtain the renewable energy output data.

[0027] Specifically, historical renewable energy output data is used as output and historical weather data is used as input to the initial output prediction model for training. The model learns the mapping relationship from weather data to renewable energy output data by adjusting the connection weights between neurons. During the training process, an optimization algorithm is used to minimize the error between the predicted value and the true value, so that the model gradually converges and the accuracy of the prediction is improved.

[0028] Preferably, in one embodiment of the present invention, the initial output prediction model is trained based on the acquired historical renewable energy output data and historical weather data, further comprising: During the training process, the hyperparameters of the initial output prediction model are optimized according to the genetic algorithm, including: A set of hyperparameter combinations is randomly generated as the initial population. Each individual in the initial population represents a set of different hyperparameter values. The initial output prediction model is used to evaluate the individual, and the fitness value of the individual is calculated according to the pre-built fitness function. The fitness value is expressed as: in, is the fitness value, , , is the weight factor; The selection, crossover and mutation operations are iteratively performed according to the fitness value until the preset number of iterations is reached to obtain the optimal individual of the current population to complete the optimization of the hyperparameters of the initial output prediction model.

[0029] Specifically, Figure 2 As shown, Figure 2 A flowchart of hyperparameter optimization of a deep neural network model provided by the present invention. Hyperparameters are parameters that need to be manually set in a deep neural network, which determine the structure and training process of the model, such as learning rate, number of hidden layers, number of neurons in each layer, activation function, etc. Different hyperparameter combinations will affect the performance and training effect of the model.

[0030] In this step, a genetic algorithm is used to optimize the selection of hyperparameters for building the neural network model, and the prediction accuracy of the model specific to power system characteristics is improved by introducing an improved fitness function.

[0031] Among them, the precision score reflects the accuracy of the model's prediction using this hyperparameter combination, which can usually be measured using indicators such as mean square error (MSE) and mean absolute error (MAE). For example, the precision score can be 1 / MSE. The smaller the MSE, the higher the precision score, indicating that the model's prediction error is smaller and the model performance is better. The computational time score takes into account the computational time required for the model to make predictions using this hyperparameter combination. The shorter the computational time, the higher the computational time score, reflecting the computational efficiency of the model. The network complexity score reflects the complexity of the model, such as the number of parameters or layers of the model. Models with too high complexity may lead to overfitting, so their scores need to be lowered.

[0032] Among them, the selection operation includes selecting excellent individuals according to the fitness value, so that excellent individuals have a higher probability of being selected to reproduce the next generation and ensure that the population evolves in a better direction. The crossover operation includes exchanging some hyperparameters of two excellent individuals to generate new individuals, increase the diversity of the population, and help find a better hyperparameter combination. For example, the learning rate of individual A and the number of hidden layers of individual B are combined into a new individual. The mutation operation includes randomly modifying the hyperparameters of individuals, changing the value of hyperparameters with a certain probability, avoiding the population from falling into the local optimum, and exploring more hyperparameter spaces.

[0033] By adding the renewable energy output data to the optimized distribution network operation model, flow calculation can be performed to obtain the flow data of each node.

[0034] Preferably, in one embodiment of the present invention, the new energy output data is input into a pre-built distribution network operation model to obtain the flow data of each node in the target distribution network, including: An initial distribution network operation model is established based on the acquired load parameters of the target distribution network; Optimizing the parameters of the target distribution network operation model according to the time series analysis method to obtain the target distribution network operation model; The renewable energy output data is input into the target distribution network operation model to obtain the flow data of each node in the target distribution network.

[0035] It is understandable that the operating parameters of the distribution network are changing dynamically, especially the load parameters, which have obvious time series characteristics. For example, the residential load shows a pattern of peaks in the morning and evening and troughs at night in a day, and the industrial load is also different on weekdays and weekends, and there may be seasonal fluctuations. The time series analysis method can capture these dynamic changes and thus optimize the parameters of the distribution network operation model.

[0036] S2: Perform sensitivity analysis on the power flow data, perform clustering based on the sensitivity analysis results, and divide the nodes into multiple initial node groups, wherein the clustering is designed to divide the nodes according to the impact of reactive power on node voltage and the electrical distance between nodes; Combine electrical distance and reactive power sensitivity to form a joint index, and perform clustering based on the joint index. Based on the clustering results, the power grid is divided into different levels, and each level independently manages its internal reactive power dispatch and control strategy.

[0037] Preferably, in one embodiment of the present invention, the sensitivity analysis result includes a sensitivity joint index, which is expressed as: in, , are node phase angle and voltage change vector respectively, , are the node active and reactive power change vectors, is the reactive power sensitivity matrix of the node, and Indicates the voltage change value corresponding to grid nodes i and j at time t; The correlation coefficient of the voltage change at node j causing the voltage change at node i is, The correlation coefficient indicating the voltage change at node i causing the voltage change at node j; represents the electrical distance between nodes i and j; is the reactive power sensitivity; is the sensitivity joint index, is the weight.

[0038] Preferably, in one embodiment of the present invention, clustering is performed based on the sensitivity analysis results to divide the nodes into a plurality of initial node groups, including: Each node is regarded as an independent cluster, and the sensitivity joint index dataset is constructed based on the sensitivity joint index of all clusters; In the process of node merging, the two clusters with the smallest distance between the sensitivity joint index data sets are merged to obtain the current sensitivity joint index data set; The node merging process is iterated until the number of clusters of the preset number of items is met to divide the nodes into multiple initial node groups.

[0039] For details, see Figure 3 , Figure 3 A distribution network hierarchical processing flow chart provided by the present invention is shown. In each iteration, two clusters with the smallest distance in the sensitivity joint index data set are found and merged. The distance here is a measurement based on the sensitivity joint index, which can be the Euclidean distance or other suitable distance measurement methods. After merging the two clusters, the current sensitivity joint index data set needs to be updated. The sensitivity joint index of the new cluster may be calculated in some way, for example, it can be the weighted average of the sensitivity joint index of the two clusters before the merger, or it can be recalculated based on the nodes of the new cluster after the merger. The above node merging process is repeated until the number of clusters that meets the preset number of items. This is a bottom-up agglomerative clustering process, which gradually divides the nodes into different groups by continuously merging similar clusters.

[0040] The multiple initial node groups finally formed can divide the power grid into different levels. The division of these levels is based on the similarity and correlation between nodes. Through sensitivity analysis and clustering operations, nodes in the same level have high similarity in terms of the impact of reactive power on voltage and electrical distance.

[0041] It is understandable that each level can independently manage its internal reactive power dispatch and control strategy. Because nodes within the same level have commonalities in their responses to reactive power changes and electrical connection relationships, it will be more effective and convenient to uniformly dispatch and control reactive power for them. For example, for nodes within a level, if they are highly sensitive to reactive power and have a close electrical distance, when reactive power imbalance occurs within the level, similar reactive compensation equipment (such as capacitor banks, static VAR compensators, etc.) can be used for adjustment, and different levels may use different reactive power adjustment methods and equipment according to their own characteristics, thereby achieving more targeted and efficient reactive power management and improving the voltage stability and operating efficiency of the power grid.

[0042] S3: Based on the acquired topological structure information of the target distribution network, the initial node group is modified to obtain multiple target node groups matching the actual physical structure; Preferably, in one embodiment of the present invention, the initial node group is modified based on the acquired topological structure information of the target distribution network to obtain multiple target node groups matching the actual physical structure, including: Obtaining topological information of the target distribution network, the topological information including node data, line connection relationship data and electrical area division data; The nodes in the initial node group are adjusted according to the topological structure information to obtain multiple target node groups that match the actual physical structure.

[0043] Among them, the node data includes the basic information of each node in the distribution network, such as the node number, location (which can be geographical coordinates or logical location in the power grid), node type (such as power node, load node, contact node, etc.), node rated voltage, node capacity, etc. The line connection relationship data describes the connection method between nodes, including which nodes are connected by lines, and the basic parameters of the lines. The electrical area division data is the regional division of nodes according to the electrical characteristics of the distribution network. For example, the distribution network can be divided into different areas according to the voltage level, power supply range or electrical coupling relationship. Different areas may have different operating characteristics. For example, high voltage level areas and low voltage level areas may differ in equipment selection, protection configuration, etc.

[0044] It is understandable that the initial node group is formed based on other factors (such as sensitivity analysis of power flow data), but may not fully consider the actual physical structure of the distribution network. The nodes in the initial node group are reallocated according to the node location information in the topology information. For example, if two nodes are divided together in the initial node group, but they are far apart in actual physical location and there are multiple other nodes in between or connected by long lines, they can be reallocated into different groups according to the topology information, so that the division of node groups is more in line with the actual geographical distribution or logical distribution.

[0045] Specifically, check whether the connection relationship between the nodes in the initial node group is consistent with the actual line connection relationship. If in an initial node group, some nodes do not have a direct physical connection or the connection line between them has a high resistance or reactance, while other nodes are more tightly connected, then the node group can be adjusted according to the line connection relationship in the topology information. For a node group, if the connection relationship between its internal nodes does not conform to the actual line connection relationship, the control strategy may not be effectively implemented. For example, when performing reactive power compensation, if the connection between the nodes in the group is loose, the expected reasonable distribution of reactive power and voltage regulation effect may not be achieved.

[0046] According to the electrical area division data, the node group is adjusted to ensure that the nodes in the same node group are in the same electrical area or have similar electrical area characteristics. Nodes in different electrical areas may have different demands and responses for reactive power, voltage regulation, etc. Dividing nodes in different electrical areas into different target node groups is conducive to formulating corresponding control strategies based on the characteristics of different areas. For example, nodes in high-voltage areas may pay more attention to voltage stability and power quality, while nodes in low-voltage areas may pay more attention to voltage levels and normal power supply of loads. By adjusting the node group, the division of the target node group is more in line with the characteristics of these electrical areas, which facilitates differentiated management and control of nodes in different areas.

[0047] S4: Input the power flow data into the pre-built mobile emergency power dispatch model and hybrid intelligent soft switch control model, and generate a primary control plan for the target distribution network according to the output results; In one embodiment of the present invention, the optimization goal is to minimize the power purchase cost of the distribution network, the load power outage compensation cost and the dispatchable resource operation cost, and consider the layout and operation constraints of the mobile emergency power supply to the access point, realize the dynamic reconstruction of the distribution network, and establish a real-time dispatch strategy for the mobile emergency power supply: dynamically adjust the layout of the mobile emergency power supply according to the grid load, photovoltaic power generation forecast data and real-time voltage fluctuations; consider the path length, node load priority and reactive power demand, and optimize the power distribution. Its objective function is: In the formula, For Node The power of the power outage load at time t; and are the substation power and mobile emergency power supply power at time t respectively; For Node Load outage compensation costs; and They are the price weight of interaction between distribution network and upper-level power grid and the operating cost weight of mobile emergency power supply respectively.

[0048] Mobile emergency power supply operation constraints: In the formula, The value is 0 or 1. When it is 1, it means that node i is connected to the mobile emergency power supply at time t. When it is 0, it means that node i is not connected to the mobile emergency power supply at time t; The number of mobile emergency power sources available for dispatch; and They are the minimum and maximum active powers of the mobile emergency power supply respectively; and They are the minimum and maximum reactive power of the mobile emergency power supply respectively; and are the active and reactive powers of the emergency power supply at node i at time t, respectively; Indicates the time when the mobile emergency power source arrives at node i.

[0049] The cost of purchasing electricity is related to the power of the substation and the corresponding price weight, reflecting the economic cost of purchasing electricity from the upper power grid. The load outage compensation cost takes into account the economic compensation caused by power outages to users, prompting the system to minimize power outages. The operating cost of the mobile emergency power supply is related to the use of the mobile emergency power supply and its cost weight, aiming to optimize the use of the mobile emergency power supply to reduce costs. By comprehensively considering these costs, the established objective function provides an economic optimization direction for the dispatch of mobile emergency power supplies, so that the distribution network can meet the power supply demand while minimizing various costs and improving the economy and reliability of the distribution network.

[0050] In one embodiment of the present invention, a smart soft switch (SOP) is connected in parallel with a traditional tie switch (TS) to form a hybrid smart soft switch to meet the dynamic capacity increase requirements of the branch. Figure 4shown.

[0051] The specific model of hybrid intelligent soft switch is as follows: SOP is the active power flowing from node i to node j: The SOP and the tie switch power need to be in the same direction (to avoid loop current) and flow in one direction: SOP active power balance equation: VSC power loss equation: SOP reactive power constraint: SOP active and reactive power injection is limited by capacity: The active power flowing from node i to node j through the tie switch is: The SOP and the tie switch power need to be in the same direction (to avoid loop current) and flow in one direction: The active and reactive power injection of the tie switch is limited by capacity: in, is the active power flowing through SOP on branch ij; , , , , , , and is the active and reactive power injected by the VSC at nodes i and j; A 0-1 variable to ensure unidirectional flow of branch power; Plan capacity for VSC; , , , is the VSC power loss at nodes i and j; , is the VSC loss coefficient at nodes i and j; , , , The upper and lower limits of VSC reactive power injection at nodes i and j; , are respectively the active and reactive power flowing through the tie switch on branch ij; , are the active powers of the tie switches at nodes i and j respectively; is the capacity of the contact switch.

[0052] Furthermore, with the goal of minimizing the annualized comprehensive cost of the distribution network, an optimization control method for hybrid intelligent soft switches is proposed. According to the calculation results of the optimization model, the transmission power of the SOP and the interconnecting switch is dynamically adjusted. The SOP responds to the global power allocation demand, and the interconnecting switch switches the operating mode.

[0053] The planning model aims to minimize the annualized comprehensive cost of the distribution network. The comprehensive cost includes operating cost and loss cost, among which the operating cost includes SOP operating cost and carbon emission cost; the loss cost includes network loss and VSC loss; a, b, c, d are weights. The objective function is expressed as follows: Running costs: 1) Hybrid intelligent soft switch operating cost: In the formula, is the SOP annual operation coefficient.

[0054] 2) Carbon emission costs: In the formula, is the carbon emission intensity per kilowatt-hour of the upper substation; is the daily emission cost.

[0055] Loss cost: 1) Network loss cost: In the formula, for Branch resistance; is the system rated voltage; , is the active and reactive power of the ij branch at time t; is the daily network loss cost, A collection of branches.

[0056] 2) SOP loss cost: In the formula, The power consumption cost of VSC inverter and DC-DC converter is shown in Figure 2.

[0057] Furthermore, the model needs to satisfy the power flow equation and safety constraints as follows: Linearized DistFlow power flow equation: 1) Contact line: In the formula, for Branch reactance; , for Active and reactive power losses of branches; , for Node active and reactive power injection, Nodes represent nodes other than the balancing node; , is the active and reactive power of the load; , Node , Voltage at time t and scenario s; The reactive power flowing through the tie switch.

[0058] 2) Other branches: In the formula, Provide active power for distributed photovoltaics.

[0059] 3) Network operation constraints: in, , are the upper and lower limits of node voltage; It is the upper limit of branch active power.

[0060] 4) PV operation constraints: in, It is the maximum value of the available active power output of the photovoltaic unit.

[0061] Preferably, in one embodiment of the present invention, the output results include grid load, photovoltaic power generation forecast data, real-time voltage fluctuations, operating costs and loss costs; The power flow data is input into the pre-built mobile emergency power dispatch model and hybrid intelligent soft switch control model respectively, and the primary control scheme of the target distribution network is generated according to the output results, including: Dynamically adjust the location and power distribution of mobile emergency power supplies based on grid load, photovoltaic power generation forecast data and real-time voltage fluctuations; And dynamically adjust the transmission power of intelligent soft switches and tie switches according to operating costs and loss costs.

[0062] It is understandable that when the grid load is high and the photovoltaic power generation forecast data shows that the photovoltaic power generation is insufficient, and the real-time voltage fluctuations indicate that the voltage of some nodes has a downward trend, it is necessary to adjust the distribution of the mobile emergency power supply and place it near the load center or the area with lower voltage to provide additional power support. The power output of the mobile emergency power supply is dynamically adjusted according to the difference between the grid load and the photovoltaic power generation forecast data and the real-time voltage fluctuations. If the grid load is greater than the photovoltaic power generation and the voltage has a downward trend, the power output of the mobile emergency power supply needs to be increased; conversely, when the photovoltaic power generation is greater than the grid load and the voltage has an upward trend, the power output of the mobile emergency power supply can be appropriately reduced, and it can even be used as an energy storage device for charging.

[0063] Intelligent soft switches can control the on and off of lines and the transmission of power. According to the analysis of operating costs and loss costs, the state of the intelligent soft switch can be adjusted to optimize the power transmission path and reduce power loss. For example, when the loss cost of a certain line is high, part of the power is transferred to other lines with lower losses through the intelligent soft switch, and the line resistance loss is reduced by changing the power transmission path. Tie switches are usually used to connect different distribution network areas. When considering operating costs and loss costs, the transmission power of tie switches is dynamically adjusted according to the grid load, voltage level and line loss in different areas. For example, when the voltage and load of two adjacent areas are unbalanced, the power is transmitted from the area with high voltage and low load to the area with low voltage and high load through the tie switch. At the same time, the cost changes brought about by such operations should be considered to ensure the reduction of overall costs.

[0064] S5: Calculate the reactive power demand of the target node group according to the power flow data and the real-time voltage data of the target distribution network, and generate a secondary control plan of the target distribution network according to the reactive power demand; Preferably, in one embodiment of the present invention, a secondary control scheme for generating a target distribution network according to reactive power demand includes: Establish control targets according to reactive power demand and formulate control plans corresponding to the control targets; The control scheme is iteratively optimized based on the particle swarm optimization algorithm to obtain the secondary control scheme of the target distribution network.

[0065] Specifically, by analyzing the power flow data and real-time voltage data, the reactive power balance state of each target node group can be determined. If the voltage in the node group is low, it may be due to insufficient reactive power, and the reactive power supply needs to be increased; conversely, if the voltage is high, it may be due to excess reactive power, and the reactive power supply needs to be reduced. The specific calculation can be based on the relationship formula between reactive power and voltage or a more complex power flow equation.

[0066] According to the calculated reactive power demand, establish a clear control target. According to the reactive power demand, determine the type and capacity of reactive compensation equipment that needs to be invested. For node groups with insufficient reactive power, you can consider investing in capacitor banks, static VAR compensators (SVCs) or static synchronous compensators (STATCOMs) and other equipment; for node groups with excess reactive power, you can adjust the operating status of these devices, such as reducing the investment in capacitor banks or adjusting the output of SVCs. If there are distributed power sources in the distribution network, and the distributed power sources have reactive power regulation capabilities (such as inverter-type distributed power sources), the reactive power output of the distributed power sources can be adjusted according to the reactive power demand to assist in the balance of reactive power.

[0067] Among them, the particle swarm optimization algorithm is a heuristic optimization algorithm that simulates the foraging behavior of bird flocks or fish schools. In the optimization of control schemes, each particle represents a control scheme, and its position can represent the parameters in the control scheme (such as the capacity of the reactive compensation equipment, the position of the transformer tap, etc.).

[0068] S6: Implement the primary control plan and the secondary control plan.

[0069] Specifically, in the process of implementing the primary and secondary control schemes, it is necessary to monitor various parameters of the distribution network in real time, including voltage, current, power, frequency, power factor, etc. By installing monitoring equipment at key nodes of the distribution network, such as voltage transformers, current transformers, power meters, etc., the monitoring data is transmitted to the monitoring center. Using technologies such as advanced measurement systems (AMI) and supervisory control and data acquisition systems (SCADA), the entire distribution network is fully monitored to timely grasp the operating status of the distribution network during the implementation of the control scheme.

[0070] Based on the monitoring data, the execution effect of the control plan is evaluated. If it is found that the expected effect is not achieved after the implementation of the control plan, or new problems arise (such as voltage exceeding the limit, current exceeding the limit, equipment overload, etc.), it is necessary to promptly feedback to the control center. According to the feedback information, the control plan is adjusted or the control strategy is re-formulated. For example, if it is found during the execution process that the voltage in a certain area still fluctuates greatly, it may be necessary to further adjust the parameters of the reactive power compensation equipment or the power distribution of the mobile emergency power supply; if the equipment is overloaded, it may be necessary to readjust the power transmission path or reduce the power demand of certain nodes.

[0071] Another embodiment of the present invention provides a distribution network control system. For details, see Figure 5 , Figure 5 The figure shows a schematic diagram of a distribution network control system in one embodiment of the present invention, which includes: An acquisition module 11 is used to acquire the predicted new energy output data of the target distribution network, input the new energy output data into a pre-built distribution network operation model, and obtain the power flow data of each node in the target distribution network; An analysis module 12 is used to perform sensitivity analysis on the power flow data, perform clustering based on the sensitivity analysis results, and divide the nodes into a plurality of initial node groups, wherein the clustering is designed to divide the nodes according to the influence of reactive power on the node voltage and the electrical distance between the nodes; A correction module 13 is used to correct the initial node group based on the acquired topological structure information of the target distribution network to obtain multiple target node groups that match the actual physical structure; The first control module 14 is used to input the power flow data into the pre-built mobile emergency power supply scheduling model and the hybrid intelligent soft switch control model, and generate a primary control plan for the target distribution network according to the output results; The second control module 15 is used to calculate the reactive power demand of the target node group according to the power flow data and the real-time voltage data of the target distribution network, and generate a secondary control scheme of the target distribution network according to the reactive power demand; The execution module 16 is used to execute the primary control scheme and the secondary control scheme.

[0072] Compared with the prior art, the embodiments of the present invention have the following advantages: 1) The present invention obtains the predicted target distribution network new energy output data and inputs it into the distribution network operation model to obtain the flow data of each node, so as to more accurately grasp the operation status of the distribution network. The uncertainty of new energy output is an important factor affecting the operation of the distribution network. Obtaining and considering its output data in advance can make the flow calculation more in line with the actual operation situation, provide an accurate data basis for subsequent regulation, and thus improve the overall efficiency and reliability of the distribution network operation.

[0073] 2) This invention performs sensitivity analysis on power flow data and clusters and divides node groups according to the impact of reactive power on node voltage and the electrical distance between nodes, which helps to gain a deeper understanding of the distribution and transmission characteristics of reactive power in the distribution network. The reactive power control scheme formulated based on this is more scientific and reasonable, which can effectively reduce the unreasonable flow of reactive power, reduce network losses, improve voltage stability, and further enhance the reliability and economy of distribution network operation.

[0074] 3) The present invention introduces a mobile emergency power supply scheduling model and a hybrid intelligent soft switch control model to generate a primary control plan, and generates a secondary control plan according to reactive power demand, providing a multi-level and multi-means control strategy for the distribution network. The mobile emergency power supply and hybrid intelligent soft switch are flexible and responsive, and can quickly adjust the power flow distribution and voltage level of the distribution network under different operating conditions, such as sudden changes in new energy output and load peak and valley changes, effectively alleviating the impact of new energy access on the distribution network and ensuring the stable operation of the distribution network.

[0075] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A distribution network control method, characterized in that: include: Acquire predicted new energy output data of the target distribution network, input the new energy output data into a pre-built distribution network operation model, and obtain power flow data of each node in the target distribution network; Performing sensitivity analysis on the power flow data, performing clustering based on the sensitivity analysis result, and dividing the nodes into a plurality of initial node groups, wherein the clustering is designed to perform node division according to the influence of reactive power on node voltage and the electrical distance between nodes; Based on the acquired topological structure information of the target distribution network, the initial node group is modified to obtain a plurality of target node groups matching the actual physical structure; The power flow data are respectively input into a pre-built mobile emergency power supply dispatching model and a hybrid intelligent soft switch control model, and a primary control scheme of the target distribution network is generated according to the output results; Calculating the reactive power demand of the target node group according to the power flow data and the real-time voltage data of the target distribution network, and generating a secondary control scheme of the target distribution network according to the reactive power demand; Execute the primary control scheme and the secondary control scheme.

2. The distribution network control method according to claim 1, characterized in that: The obtaining of predicted new energy output data of the target distribution network includes: Construct an initial output prediction model based on deep neural network; The initial output prediction model is trained based on the acquired historical renewable energy output data and historical weather data to obtain a trained target output prediction model; In the process of predicting the new energy output of the target distribution network, real-time weather data is input into the target output prediction model to obtain the new energy output data.

3. The distribution network control method according to claim 2, characterized in that: The training of the initial output prediction model based on the acquired historical renewable energy output data and historical weather data further includes: During the training process, the hyperparameters for constructing the initial output prediction model are optimized according to the genetic algorithm, specifically including: Randomly generate a set of hyperparameter combinations as an initial population, where each individual in the initial population represents a set of different hyperparameter values; The individual is evaluated using the initial output prediction model, and the fitness value of the individual is calculated according to the pre-constructed fitness function. The fitness value is expressed as: in, is the fitness value, , , is the weight factor; The selection, crossover and mutation operations are iteratively performed according to the fitness value until a preset number of iterations is reached to obtain the optimal individual of the current population, so as to complete the optimization of the hyperparameters of the initial output prediction model.

4. The distribution network control method according to claim 1, characterized in that: The step of inputting the new energy output data into a pre-built distribution network operation model to obtain the flow data of each node in the target distribution network includes: Establishing an initial distribution network operation model based on the acquired load parameters of the target distribution network; Optimizing the parameters of the target distribution network operation model according to a time series analysis method to obtain the target distribution network operation model; The new energy output data is input into the target distribution network operation model to obtain the flow data of each node in the target distribution network.

5. The distribution network control method according to claim 1, characterized in that: The sensitivity analysis results include a sensitivity joint index, expressed as: in, , are node phase angle and voltage change vector respectively, , are the node active and reactive power change vectors, is the reactive power sensitivity matrix of the node, and Indicates the voltage change value corresponding to grid nodes i and j at time t; The correlation coefficient of the voltage change at node j causing the voltage change at node i is, The correlation coefficient indicating the voltage change at node i causing the voltage change at node j; represents the electrical distance between nodes i and j; is the reactive power sensitivity; is the sensitivity joint index, is the weight.

6. The distribution network control method according to claim 1, characterized in that: The step of modifying the initial node group based on the acquired topological structure information of the target distribution network to obtain multiple target node groups matching the actual physical structure includes: Acquire topological structure information of the target distribution network, wherein the topological structure information includes node data, line connection relationship data and electrical area division data; The nodes in the initial node group are adjusted according to the topological structure information to obtain a plurality of target node groups matching the actual physical structure.

7. The distribution network control method according to claim 1, characterized in that: The clustering is performed based on the sensitivity analysis result to divide the nodes into a plurality of initial node groups, including: Each node is regarded as an independent cluster, and the sensitivity joint index dataset is constructed based on the sensitivity joint index of all clusters; In the node merging process, two clusters with the smallest distance between the sensitivity joint indicator data sets are merged to obtain the current sensitivity joint indicator data set; The node merging process is iterated until a preset number of clusters is met, so as to divide the nodes into a plurality of initial node groups.

8. The distribution network control method according to claim 1, characterized in that: The output results include grid load, photovoltaic power generation forecast data, real-time voltage fluctuations, operating costs and loss costs; The power flow data is respectively input into the pre-built mobile emergency power dispatch model and the hybrid intelligent soft switch control model, and a primary control scheme of the target distribution network is generated according to the output results, including: Dynamically adjusting the location and power distribution of the mobile emergency power supply according to the grid load, the photovoltaic power generation forecast data and the real-time voltage fluctuation; And dynamically adjust the transmission power of the intelligent soft switch and the tie switch according to the operating cost and loss cost.

9. The distribution network control method according to claim 1, characterized in that: The secondary control scheme for generating the target distribution network according to the reactive power demand includes: Establishing a control target according to the reactive power demand, and formulating a control plan corresponding to the control target; The control scheme is iteratively optimized based on a particle swarm optimization algorithm to obtain a secondary control scheme for the target distribution network.

10. A distribution network control system, characterized in that: include: An acquisition module is used to acquire predicted new energy output data of a target distribution network, input the new energy output data into a pre-built distribution network operation model, and obtain power flow data of each node in the target distribution network; An analysis module, configured to perform sensitivity analysis on the power flow data, perform clustering based on the sensitivity analysis result, and divide the nodes into a plurality of initial node groups, wherein the clustering is designed to perform node division based on the influence of reactive power on node voltage and the electrical distance between nodes; A correction module, used to correct the initial node group based on the acquired topological structure information of the target distribution network to obtain multiple target node groups matching the actual physical structure; A first control module is used to input the power flow data into a pre-built mobile emergency power supply scheduling model and a hybrid intelligent soft switch control model, and generate a primary control scheme of the target distribution network according to the output results; A second control module is used to calculate the reactive power demand of the target node group according to the power flow data and the real-time voltage data of the target distribution network, and generate a secondary control scheme of the target distribution network according to the reactive power demand; An execution module is used to execute the primary control scheme and the secondary control scheme.

Images