A method for constructing a power distribution network multi-scenario joint voltage control model, a voltage control method and system, an electronic device, and a storage medium

Abstract

The application provides a power distribution network multi-scene joint voltage control model construction method, a voltage control method, a system, an electronic device, a storage medium, and relates to the technical field of power distribution network voltage control; the joint voltage control model construction method comprises the following steps: obtaining power characteristic data at time t1, determining whether a new energy station needs to participate in voltage control; if so, obtaining historical power flow data, including basic data and historical state space data; selecting and constructing a new energy station group based on the basic data; sending the historical state space data to the new energy station group for model construction and training; receiving an encrypted voltage control model; decrypting the encrypted voltage control model to obtain a decrypted voltage control model; performing parameter optimization on the decrypted voltage control model to obtain a joint voltage control model; and having the beneficial effect of enabling collaborative control among the new energy station group and realizing safe, stable and reliable operation of the power distribution network multi-scene joint voltage control.

Description

Technical Field

[0001] This application relates to the technical field of voltage control in distribution networks, specifically to a method for constructing a multi-scenario joint voltage control model for distribution networks, a voltage control method, a system and electronic equipment, and a storage medium. Background Technology

[0002] With the rapid development of new power systems, higher requirements are being placed on voltage stability, safety, and reliability. Traditional voltage control strategies and methods are no longer able to meet the needs of new power systems. In particular, the integration of a large number of distributed new energy power plants in the distribution network has brought great challenges to the voltage control of the distribution network.

[0003] Traditional distribution network voltage control methods rely on the Automatic Voltage Control (AVC) system, which mainly controls the switching of grid capacitors and the adjustment of main transformer taps based on the voltage of substations at each voltage level and the reactive power flow, in order to achieve reactive power balance and voltage stability among substations at each voltage level in the distribution network. In contrast, the voltage control of generator sets mainly adjusts the power of synchronous generators based on the power factor, in order to achieve voltage stability of the grid bus and control of active power.

[0004] After large-scale distributed renewable energy power plants are connected to the distribution network, they alter the traditional power flow direction and profoundly affect voltage variation patterns. Traditional distribution network voltage control methods alone cannot meet the needs of this new power system, primarily in two aspects: First, there are significant differences between the voltage control modes of renewable energy power plants and traditional generator sets. Traditional generator sets can respond to changes in reactive power and voltage based on the power factor. However, renewable energy power plants mainly consist of a certain number of photovoltaic panels connected to collector lines for grid-connected power generation. Each renewable energy power plant has several collector lines. To maximize the benefits of photovoltaic power generation, the collector line inverters often set the power factor to 0.95~1 through parameter control, thereby maximizing the output of the collector lines. However, the power factor of renewable energy power plants can only reflect changes in the power level at the grid connection point and cannot provide an effective basis for control. Second, due to the significant variations in renewable energy generation characteristics, external environment, wind and solar power restriction policies, and market regulation, the voltage control of renewable energy power plants cannot maintain a stable power generation level like traditional generator sets.

[0005] To address the aforementioned issues, current renewable energy power plants utilize station AVC (Automatic Voltage Control) for voltage control, primarily divided into two modes: station-wide voltage control and station-wide reactive power control. Station-wide voltage control uses the grid connection point voltage as the control target, calculates the total reactive power required for the entire station based on the target voltage value, and distributes this total reactive power according to a predetermined allocation strategy to each inverter or adjustable reactive power device participating in AVC control. This ensures the grid connection point bus voltage reaches the target value, achieving automatic voltage and reactive power control for the entire station. Station-wide reactive power control uses the total reactive power at the grid connection point as the control target, distributing it according to a predetermined allocation strategy to each inverter or adjustable reactive power device participating in AVC control, ensuring the total reactive power for the entire station reaches the target value. This achieves automatic voltage and reactive power control for all photovoltaic power generation units or adjustable reactive power devices in the entire station. Typically, renewable energy power plants primarily use a station-wide voltage control mode, where voltage control is performed based on the grid-connected bus voltage curve issued by the dispatch master station. The station-wide reactive power control mode relies on the calculation of the station's reactive power demand, which involves not only changes in the station's own impedance characteristics but also the impact of the impedance characteristics of the power grid and adjacent renewable energy power plants. Therefore, the control mechanism is more complex. At the same time, the purpose of the station-wide reactive power control mode is voltage stability, while the station-wide voltage control mode can directly reflect the level of voltage change and effectively achieve the voltage control objective.

[0006] However, in the actual operation of new energy power plants, three problems have been found in the current whole-station voltage control mode: 1. When the new energy power plant uses the grid-connected bus voltage curve as the control target, in order to maximize the output of new energy, the reactive power of the collector is controlled within a small range, so that the power factor is close to 1. The reactive power compensation device (SVG) will reserve a certain reactive power margin to provide reasonable adjustment space for the power plant. Therefore, when performing voltage control, a large amount of reactive power will be absorbed from the main grid for reactive power compensation of the power plant, resulting in frequent flow of reactive power between the grid and increasing grid losses. At the same time, the large reactive power absorption of the main grid substation will lead to insufficient reactive power margin and insufficient support for safe and stable operation; 2. The AV of the new energy power plant C. The interaction between C and the main grid AVC is insufficient, relying solely on the voltage control curves issued by the main station for voltage control of lower-level substations. It cannot involve the control of reactive power regulation equipment such as substation collector inverters and SVG. As a result, when the reactive power reserve of the substation is insufficient, it is impossible to mobilize the collector inverters and SVG of the renewable energy substations to provide reactive power support to the main grid, affecting the safe and stable operation of the power grid. 3. When multiple renewable energy substations participate in voltage control within the same 220kV substation zone, there is a lack of a unified and coordinated control strategy to achieve voltage stability and local reactive power balance. It is necessary to consider not only the safety and stability of the power grid, but also the safety and stability of each renewable energy substation, which places higher demands on the participation of renewable energy substations in distribution network voltage control. Summary of the Invention

[0007] To address one of the aforementioned technical deficiencies, this application provides a method for constructing a multi-scenario joint voltage control model for power distribution networks, a voltage control method, a system and electronic equipment, and a storage medium.

[0008] According to the first aspect of this application, a method for constructing a multi-scenario joint voltage control model for a distribution network is provided, including:

[0009] Obtain the power characteristic data at time t1, and determine whether the new energy power station needs to participate in voltage control based on the power characteristic data;

[0010] If the renewable energy power station needs to participate in voltage control, then perform the following operations:

[0011] Acquire historical trend data, which includes basic data and historical state space data;

[0012] Based on basic data, select renewable energy power stations that can participate in voltage regulation from all renewable energy power stations and construct a group of renewable energy power stations;

[0013] Historical state space data is sent to a group of new energy power stations for model building and training;

[0014] Receive encrypted voltage control models trained based on historical state space data from each renewable energy power station that can participate in voltage regulation, and record the number of times the models are received.

[0015] The encrypted voltage control model is decrypted to obtain the decrypted voltage control model.

[0016] The parameters of the decrypted voltage control model are optimized to obtain a joint voltage control model.

[0017] According to a second aspect of this application, a multi-scenario joint voltage control method for a distribution network is provided, comprising the following steps:

[0018] Collect state-space data at the current moment;

[0019] Input the current state space data into the joint voltage control model constructed using the aforementioned method for constructing a multi-scenario joint voltage control model for distribution networks;

[0020] The action space vector at the current moment is output through the joint voltage control model;

[0021] Collect the second basic data at the current moment, calculate the reactive power of all collectors and all SVGs in the new energy power plants that can participate in voltage regulation based on the second basic data at the current moment, and execute the action space vector at the current moment according to the ratio of the reactive power of all collectors and the reactive power of all SVGs.

[0022] According to a third aspect of this application, a multi-scenario joint voltage control system for a distribution network is provided, including a module for implementing the aforementioned method for constructing a multi-scenario joint voltage control model for a distribution network, or a module for implementing the aforementioned method for multi-scenario joint voltage control of a distribution network.

[0023] According to a fourth aspect of this application, an electronic device is provided, comprising:

[0024] Memory;

[0025] Processor; and

[0026] Computer programs;

[0027] The computer program is stored in the memory and configured to be executed by the processor to implement the aforementioned method for constructing a multi-scenario joint voltage control model for a distribution network, or to implement the aforementioned method for joint voltage control of a distribution network in multiple scenarios.

[0028] According to a fifth aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon; the computer program is executed by a processor to implement the aforementioned method for constructing a multi-scenario joint voltage control model for a power distribution network, or is executed to implement the aforementioned method for joint voltage control of a power distribution network in multiple scenarios.

[0029] The beneficial effects of this application are as follows:

[0030] Based on the above scheme, a determination was made regarding the need for renewable energy power plants to participate in voltage control. When participation was required, renewable energy power plants eligible for voltage regulation were selected using historical power flow data and formed a renewable energy power plant group. Historical state space data was sent to this group to train the model within the renewable energy power plants. Upon receiving the trained encrypted voltage control model, it was decrypted and its parameters optimized to obtain a joint voltage control model. This enables collaborative control among the renewable energy power plant group, achieving safe, stable, and reliable operation of the distribution network's multi-scenario joint voltage control. This avoids control conflicts caused by the relatively independent voltage control strategies of renewable energy power plants and the distribution network, effectively realizing voltage collaborative control between the power source and the grid. It provides a more scientific, reasonable, efficient, and precise guidance strategy and decision-making for multi-scenario joint voltage control of the distribution network.

[0031] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of what is pointed out in the written description and the accompanying drawings. Attached Figure Description

[0032] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0033] Figure 1 A flowchart illustrating the construction method of a multi-scenario joint voltage control model for a power distribution network provided in this application;

[0034] Figure 2 A flowchart illustrating the process for determining whether a new energy power station needs to participate in voltage control, provided for this application;

[0035] Figure 3 A flowchart illustrating a multi-scenario joint voltage control method for a power distribution network provided in this application;

[0036] Figure 4 This application provides a schematic diagram of the functional module structure of a multi-scenario joint voltage control system for power distribution networks.

[0037] Figure 5 for Figure 4 A schematic diagram of the functional module structure of the judgment module;

[0038] Figure 6 for Figure 4 A schematic diagram of the functional module structure of the selected unit;

[0039] Figure 7 for Figure 6 A schematic diagram of the functional module structure of the reactive power calculation unit in China;

[0040] Figure 8 This application provides a schematic diagram of another functional module structure for a multi-scenario joint voltage control system for power distribution networks.

[0041] Figure 9 for Figure 8 A schematic diagram of the functional module structure of the computing module;

[0042] In the picture:

[0043] 10 is the acquisition module, 20 is the judgment module, 30 is the execution module, 40 is the first acquisition module, 50 is the input module, 60 is the output module, 70 is the second acquisition module, 80 is the calculation module, 90 is the execution voltage control module, 201 is the first judgment unit, 202 is the second judgment unit, 203 is the determination unit, 301 is the acquisition unit, 302 is the selection unit, 303 is the sending unit, 304 is the receiving unit, 305 is the decryption unit, 306 is the parameter optimization unit, and 801 is the third calculation unit. 802 is the current scene and formula acquisition unit, 803 is the fourth calculation unit, 3021 is the sensitivity matrix calculation unit, 3022 is the sensitivity matrix processing unit, 3023 is the reactive power calculation unit, 3024 is the total electrical feature vector generation unit, 3025 is the aggregation unit, 3026 is the node feature matrix construction unit, 3027 is the transformation unit, 3028 is the classification unit, 30231 is the first calculation unit, 30232 is the scene and formula acquisition unit, and 30233 is the second calculation unit. Detailed Implementation

[0044] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0045] To address the aforementioned problems, according to the first aspect of this application, a method for constructing a multi-scenario joint voltage control model for a distribution network is provided, comprising:

[0046] Obtain the power characteristic data at time t1, and determine whether the new energy power station needs to participate in voltage control based on the power characteristic data;

[0047] If the renewable energy power station needs to participate in voltage control, then perform the following operations:

[0048] Acquire historical trend data, which includes basic data and historical state space data;

[0049] Based on basic data, select renewable energy power stations that can participate in voltage regulation from all renewable energy power stations and construct a group of renewable energy power stations;

[0050] Historical state space data is sent to the new energy power plant group for model building and training. The historical state space data includes: substation bus voltage, bus voltage of all new energy power plants that can participate in voltage regulation, reactive power of substation, and reactive power exchanged between all new energy power plants that can participate in voltage regulation and substation.

[0051] The system receives encrypted voltage control models trained based on historical state space data from various renewable energy power stations that can participate in voltage regulation, and records the number of receptions (i.e., the number of communications).

[0052] The encrypted voltage control model is decrypted to obtain the decrypted voltage control model.

[0053] The parameters of the decrypted voltage control model are optimized to obtain a joint voltage control model.

[0054] Based on the above scheme, a determination was made regarding the need for renewable energy power plants to participate in voltage control. When participation was required, renewable energy power plants eligible for voltage regulation were selected using historical power flow data and formed a renewable energy power plant group. Historical state space data was sent to this group to train the model within the renewable energy power plants. Upon receiving the trained encrypted voltage control model, it was decrypted and its parameters optimized to obtain a joint voltage control model. This enables collaborative control among the renewable energy power plant group, achieving safe, stable, and reliable operation of the distribution network's multi-scenario joint voltage control. This avoids control conflicts caused by the relatively independent voltage control strategies of renewable energy power plants and the distribution network, effectively realizing voltage collaborative control between the power source and the grid. It provides a more scientific, reasonable, efficient, and precise guidance strategy and decision-making for multi-scenario joint voltage control of the distribution network.

[0055] In some possible implementations of the first aspect, the power characteristic data includes the substation bus voltage, the reactive power absorbed by each new energy power plant from the substation, and the substation's reactive power static reserve coefficient.

[0056] The determination of whether renewable energy power plants need to participate in voltage control based on power characteristic data specifically includes:

[0057] Determine the static reactive power reserve coefficient of the substation at time t1 Whether it is below the coefficient threshold (which can be 0.1~0.15); where, , Let t1 be the reactive power of the substation. The critical value of reactive power when the voltage of the substation collapses can be obtained from the voltage-reactive power curve of the substation.

[0058] If it is below the coefficient threshold, that is (This indicates that the safety and stability of the substation are threatened, and the new energy power station must be called upon first for voltage control), so it is directly determined that the new energy power station needs to participate in voltage control;

[0059] If it is not lower than the coefficient threshold, then determine the bus voltage of the substation at time t1. Is it outside the preset voltage range? Within this context, does the sum of reactive power absorbed by each new energy power station from the substation at time t1 exceed the reactive power consumed by the substation at time t1? 50%;

[0060] If so, that is or ,and ,(in, Let t1 represent the reactive power absorbed by the i-th renewable energy power station from the substation, and n represent the number of centralized renewable energy power stations of 10kV and above connected to the distribution network within the substation zone. If the voltage control is required, then it is determined that the new energy power station needs to participate in the voltage control.

[0061] Otherwise, it is determined that the participation of new energy power stations in voltage control is not required.

[0062] Based on the above scheme, specific analysis and judgment were carried out on the situations in which new energy power stations need to participate in voltage control, so that the constructed joint voltage control model is more in line with the usage environment and more accurate and reliable.

[0063] In some possible implementations of the first aspect, the basic data includes first basic data and second basic data; the first basic data includes the reactive power and bus voltage of each new energy power station; the second basic data includes the compensable reactive power of each new energy power station, the active power and reactive power of all collector lines and the rated capacity of the collector line inverters, the capacity, reactive power and set reserved reactive power margin of all SVGs.

[0064] The process of selecting renewable energy power stations that can participate in voltage regulation from all renewable energy power stations based on basic data and constructing them into a renewable energy power station group specifically includes:

[0065] Calculate the voltage sensitivity matrix of the new energy power station based on the first set of basic data. For voltage sensitivity matrix Normalization and binarization are performed to obtain the relation matrix. ;

[0066] Based on the second set of basic data, the reactive power generated by all collectors and the reactive power generated by all SVGs in each new energy power station are calculated, and these are combined with the bus voltage of the corresponding new energy power station to construct the electrical characteristic vector of each new energy power station. ; The electrical characteristic vectors of each new energy power station The components are merged to generate a total electrical feature vector. ;

[0067] The relation matrix is ​​analyzed using aggregation formulas. Sum of total electrical eigenvectors Perform node feature aggregation to obtain an aggregated vector. A graph attention mechanism is used to extract node features from the aggregated vector, and the extracted node features are then used to construct a node feature matrix. ;

[0068] The node feature matrix is ​​processed through a fully connected layer. Perform a linear transformation to obtain the sample matrix. ; classify the sample matrix using a classification strategy The renewable energy power station nodes are classified into two categories: those that can participate in voltage regulation (the node voltage of the renewable energy power station is sensitive and the reactive power regulation resources are sufficient) and those that do not participate in voltage regulation (the node voltage of the renewable energy power station is not sensitive and the reactive power regulation resources are insufficient). The renewable energy power stations that can participate in voltage regulation are constructed into a renewable energy power station group E. ;in, , These represent the first to the Hth renewable energy power stations that can participate in voltage regulation.

[0069] Based on the above scheme, the voltage sensitivity matrix is ​​calculated. Then, it was normalized and binarized to obtain the relation matrix. This provides a reliable foundation for voltage control and is more accurate and effective compared to traditional topology node matrices; the relationship matrix is ​​aggregated using a convergence formula. Sum of total electrical eigenvectors Node feature aggregation was performed, resulting in an aggregated vector. Then, node features are extracted using a graph attention mechanism. During this process, the relational weights of neighboring nodes are dynamically assigned based on the electrical characteristics of each new energy power station (e.g., This can effectively improve the ability of the trained joint voltage control model to capture node correlations and enhance the accuracy of the joint voltage control model.

[0070] In some possible implementations of the first aspect, the calculation of the voltage sensitivity matrix of the renewable energy power station based on the first basic data... Specifically, it includes:

[0071] Calculate the voltage sensitivity matrix based on the modified equation. The corrected equation is:

[0072] ;

[0073] In the formula, , This represents the bus voltage amplitude of the i-th renewable energy power station at time t2. , Let t2 and t2-1 represent the bus voltages of the i-th renewable energy power station, respectively. , Let represent the change in reactive power of the i-th renewable energy power station at time t2. , Let represent the reactive power of the i-th renewable energy power station at time t2 and time t2-1, respectively;

[0074] Voltage sensitivity matrix Represented as: ;

[0075] in, , This represents the sensitivity of the bus voltage amplitude of the i-th renewable energy power station at time t2 to the reactive power change of the n-th renewable energy power station.

[0076] In some possible implementations of the first aspect, the voltage sensitivity matrix Normalization and binarization are performed to obtain the relation matrix. Specifically, it includes:

[0077] Take the voltage sensitivity matrix The largest element in and minimum element ;

[0078] Maximum element Represented as: ;

[0079] minimum element Represented as: ;

[0080] Normalization is performed according to the normalization formula to obtain the normalized matrix. The normalization formula is: ;

[0081] Normalized matrix Represented as: ;

[0082] The normalized matrix is ​​processed according to the binarization formula. Binarization is performed to obtain the relation matrix. ;

[0083] The formula for binarization is: ;

[0084] In the formula, , This indicates a high sensitivity threshold, typically in the range of 35~110kV. The voltage is 0.2~0.5kV / MVar, while in this application... It is 1kV / MVar.

[0085] In some possible implementations of the first aspect, the reactive power generated by all collectors and the reactive power generated by all SVGs in each renewable energy power station are calculated based on the second basic data, specifically including:

[0086] Calculate the static reserve factor of reactive power at time t2. and the power factor of each new energy power station ;

[0087] In the formula, This represents the reactive power of the substation at time t2. The critical value of reactive power when the voltage of the substation collapses can be obtained from the voltage-reactive power curve of the substation.

[0088] In the formula, This represents the power factor of the i-th renewable energy power station at time t2. This represents the active power of all collectors in the i-th renewable energy power station at time t2. This represents the reactive power of all collectors in the i-th renewable energy power station at time t2;

[0089] Based on the reactive power static reserve coefficient at time t2 and the power factor of each new energy power station Obtain the calculation formula for the current scenario and the corresponding reactive power;

[0090] The reactive power of all collectors and all SVGs in each new energy power station at time t2 is calculated using the formula for calculating reactive power. (The calculated reactive power is the limit of adjustable reactive power and can be used to characterize the magnitude of reactive power regulation capability.)

[0091] The formula for calculating the reactive power that can be generated at time t2 is expressed as:

[0092] ;

[0093] In the formula, , This represents the reactive power adjustment range of the m-th SVG at the i-th renewable energy power station at time t2. Let m represent the capacity of the m-th SVG in the i-th renewable energy power station. This indicates the set reactive power margin of the m-th SVG at the i-th renewable energy power station at time t2. Let t2 represent the reactive power of the m-th SVG at the i-th renewable energy power station;

[0094] , Let M represent the adjustment coefficient of the m-th SVG of the i-th renewable energy power station at time t2, and M represent the number of SVGs in the i-th renewable energy power station. This represents the amount of reactive power that the i-th renewable energy power station can compensate at time t2. This value is determined by the reactive power adjustment potential of the renewable energy power station and is provided by each renewable energy power station (not the total upper limit, but the reactive power adjustment range will increase when renewable energy curtails wind and solar power or when active power control is implemented. It is obtained by the renewable energy power station after recalculating the reactive power balance within the station after active power control is implemented). Let t2 represent the reactive power generated by the k-th collector line of the i-th renewable energy power station; K represents the number of collector lines in the i-th renewable energy power station.

[0095] , Let represent the adjustment coefficients of the k-th collector line inverter of the i-th renewable energy power station at time t2 under different scenarios. , This represents the active power of the k-th collector line of the i-th renewable energy power station at time t2. It is collected and calculated by the voltage and current transformers of the power station and then uploaded to the SCADA system, where it can be directly obtained. , This represents the rated capacity of the inverter on the k-th collector line of the i-th renewable energy power station.

[0096] Since the formulas and the meaning of each letter in the calculation formula have been explained above, the meaning of each letter will not be repeated below for the sake of brevity.

[0097] In some possible implementations of the first aspect, based on the reactive static reserve coefficient at time t2 and the power factor of each new energy power station The calculation formula for obtaining the current scenario and the corresponding reactive power output includes:

[0098] Determine if the first condition is met: ;

[0099] If the first condition is not met (i.e.) If the condition is true, it indicates that the reactive power margin of the distribution network substation cannot meet the voltage stability requirements.

[0100] In this scenario, the i-th renewable energy power station needs to reduce its active power (wind and solar curtailment) and increase the range of reactive power adjustment. This requires coordinated compensation using collector inverters and SVG. The collector inverters are used for compensation first, while the remaining capacity is dynamically compensated by SVG, and the reactive power margin of the i-th renewable energy power station is preserved.

[0101] In this scenario, the formula for calculating the reactive power generated by all collectors in each renewable energy power station at time t2 is: In the formula, the adjustment coefficient of the inverter of the k-th collector line of the i-th renewable energy power station at time t2 is: ;

[0102] The adjustment coefficient of the m-th SVG of the i-th renewable energy power station at time t2 The calculation is based on the reactive power compensation of the collector-line inverter: ;

[0103] In this scenario, the formula for calculating the reactive power of all SVGs in each renewable energy power station at time t2 is: ;

[0104] If the first condition is met, then determine whether the second condition is met. If the second condition is met, it means that the scenario is one where the reactive power flowing between the substation and each new energy power station is relatively small (this may be a photovoltaic power station at night, on a cloudy day, or a wind farm in a weak wind).

[0105] In this scenario, the reactive power balance of the substation can be satisfied and the optimal reactive power distribution can be achieved by adjusting the SVG. The AVC of the new energy power station performs reactive power control according to the proportion of reactive power that each SVG can generate at time t2.

[0106] However, the adjustment range of SVG needs to consider two conditions: (1) the new energy power station has a suitable reactive power margin; (2) avoid reactive power circulation caused by the imbalance of reactive power with the collector line.

[0107] In this scenario, the formula for calculating the reactive power of all SVGs in each renewable energy power station at time t2 is: In the formula, Typically, renewable energy power plants will select one of their units to have a reserved reactive power margin (i.e., This will improve the reactive static stability of new energy power plants.

[0108] If the second condition is not met, then determine whether the third condition is met. If the third condition is met, it means that the scenario is an increase in reactive power exchange between the new energy power station and the substation (which may be the daytime mode of the photovoltaic power station, or when the wind power of the wind farm is sufficient).

[0109] In this scenario, since SVG regulation is no longer sufficient, we consider using the dispatch collector inverter for reactive power regulation. While maintaining the active power generation (power factor) of the new energy power station, we can call upon reactive power resources and achieve local reactive power balance through reactive power mutual assistance between the substation and each new energy power station.

[0110] In this scenario, the formula for calculating the reactive power generated by all SVGs in each renewable energy power station at time t2 is: ;

[0111] The formula for calculating the reactive power generated by all collectors in each renewable energy power station at time t2 is as follows: In the formula, ;

[0112] If the third condition is not met, it means that the scenario is that the new energy power station is in a period of high wind and solar power generation, and absorbs too much reactive power from the main grid substation, resulting in an increase in reactive power imbalance in the substation; usually, the new energy power station will participate in frequency regulation control, that is, the automatic generation control (AGC) system controls the reduction of active power.

[0113] In this scenario, due to the good dynamic adjustment performance of SVG, the reactive power margin of SVG is preserved, and the collector inverter is given priority for adjustment. At this time, active power reduction is required to adjust the substation voltage and reactive power balance.

[0114] In this scenario, the formula for calculating the reactive power generated by all collectors in each renewable energy power station at time t2 is: , ;

[0115] The formula for calculating the reactive power generated by all SVGs in each renewable energy power station at time t2 is as follows: .

[0116] Based on the above scheme, the scenario at time t2 is obtained by judging the first, second, and third conditions, and the calculation formula for the reactive power that can be generated under the corresponding scenario is obtained.

[0117] in:

[0118] 1. When the first condition is not met (i.e.) The scenario is that the reactive power margin of the distribution network substation cannot meet the voltage stability requirements. In this scenario, it is necessary to rely on the collector inverter and SVG for coordinated compensation. The collector inverter is used first for compensation, and the remaining capacity is dynamically compensated by the SVG, while the reactive power margin of the i-th new energy power station is reserved.

[0119] 2. When the first and second conditions are met (i.e.) The scenario involves a relatively small amount of reactive power flowing between the substation and various new energy power plants (possibly at night, on cloudy days, or in wind farms during periods of weak wind). In this scenario, the reactive power balance of the substation can be achieved through the adjustment of the SVG, thus realizing the optimal distribution of reactive power.

[0120] 3. When the first and third conditions are met (i.e.) The scenario is an increase in reactive power exchange between new energy power plants and substations (possibly during the daytime mode of photovoltaic power plants, or when wind power is sufficient in wind farms, etc.).

[0121] In this scenario, since SVG regulation is no longer sufficient, we consider using the dispatch collector inverter for reactive power regulation. While maintaining the active power generation (power factor) of the renewable energy power plant, we can utilize reactive power resources and achieve local reactive power balance through reactive power mutual assistance between the substation and each renewable energy power plant.

[0122] 4. When the first condition is met, but the second and third conditions are not met (i.e.) The scenario involves a renewable energy power plant operating during peak wind and solar power generation periods, absorbing excessive reactive power from the main grid substation, leading to increased reactive power imbalance at the substation. Typically, renewable energy power plants participate in frequency regulation control, i.e., the AGC system controls active power reduction. In this scenario, due to the excellent dynamic adjustment performance of the SVG, the reactive power margin of the SVG is preserved, and the collector-line inverter is given priority for regulation. At this time, active power reduction is required to adjust the substation voltage and reactive power balance.

[0123] In some possible implementations of the first aspect, electrical feature vectors Represented as:

[0124] ;

[0125] In the formula, This represents the bus voltage of the i-th renewable energy power station at time t2. It represents the reactive power generated by the first to the Kth collector lines of the i-th renewable energy power station at time t2 (i.e., the reactive power generated by all collector lines in the renewable energy power station). It represents the reactive power generated by the first to the Mth SVG of the i-th renewable energy power station at time t2 (i.e., the reactive power generated by all SVGs in the renewable energy power station).

[0126] Total electrical eigenvector Represented as: ;

[0127] The aggregation formula is expressed as: ;

[0128] In the formula, Represents an aggregate vector. This represents the activation function. Represents the identity matrix. express The degree matrix;

[0129] Aggregate vector Represented as: ;

[0130] The extraction formula for the graph attention mechanism is expressed as: ;

[0131] In the formula, This represents the feature vector of the i-th node (i.e., the i-th renewable energy power station) extracted by the attention mechanism at time t2. This represents the activation function. Let N represent the set of all N neighboring node features that are adjacent to the i-th node feature. This represents the relationship weight between the feature of the i-th node and the feature of its j-th neighboring node. This represents the aggregated vector of features of the j-th neighboring node at time t2;

[0132] , and This indicates a trainable weighted network, which generates a weighted network adapted to the node features based on the training process after parameter initialization. Represents a non-linear activation function layer; , , Let i and j represent the aggregate vectors of the features of the i-th node, the aggregate vectors of the features of the j-th neighboring node, and the aggregate vectors of the features of the j-th neighboring node, respectively. An aggregate vector of features from neighboring nodes. Used for counting;

[0133] Node feature matrix Represented as: ;

[0134] In the formula, Indicates the activation function;

[0135] The formula for the linear transformation is expressed as: ;

[0136] In the formula, This represents the weight matrix of the fully connected layer, and the parameter matrix representing the connections between each neuron in each layer and each neuron in the previous layer. These parameters are automatically updated and adjusted during training of the fully connected layer. This represents the bias matrix, indicating that each output neuron has an independent bias term used to adjust the overall level of the output.

[0137] The sample matrix is ​​represented as follows: ;

[0138] In the formula, This represents the sample data of the first to the nth renewable energy power stations. The dimension of the sample data is the number of categories, q. In this application, q=2 (including two categories: renewable energy power stations that can participate in voltage regulation and renewable energy power stations that do not participate in voltage regulation).

[0139] Classification strategies include the softmax classifier and the argmax algorithm, where:

[0140] The classification formula for the softmax classifier is expressed as follows:

[0141] ;

[0142] In the formula, This represents the sample data of the i-th renewable energy power station. This represents the classification result corresponding to the sample data of the i-th renewable energy power station. Let represent the probability value of each type of output result corresponding to the sample data of the i-th renewable energy power station, where These are built-in parameters for the Softmax classifier.

[0143] After calculating the probabilities using the softmax classifier, the argmax algorithm can be used to classify each renewable energy power station into renewable energy power stations that can participate in voltage regulation and those that do not.

[0144] In some possible implementations of the first aspect, the parameter optimization of the decryption voltage control model to obtain a joint voltage control model specifically includes:

[0145] The model parameters in the decrypted voltage control model are globally optimized and updated using the FedAvg algorithm.

[0146] Determine whether the number of receptions g has reached the preset number G. If the preset number has not been reached, send the globally optimized and updated model parameters to the new energy power station group for model retraining.

[0147] If the preset number of iterations is reached, the globally optimized and updated model parameters are output, resulting in a joint voltage control model.

[0148] In some possible implementations of the first aspect, the global objective function of the FedAvg algorithm is:

[0149] ;

[0150] In the formula, This indicates that after the g-th global optimization update (corresponding to the number of receptions) at time t3, the new energy power station group will... Next (at this time) The global objective function after model training. This indicates that the h-th renewable energy power station eligible for voltage regulation is activated after the g-th global optimization update at time t3. Next (at this time) The local objective function after model training. This represents the number of local data points for the h-th renewable energy power station that can participate in voltage regulation. This represents the amount of local data from all H renewable energy power stations that can participate in voltage regulation (the renewable energy power stations that contribute more data in each round have a greater training weight in the global optimization update). G represents the preset number of times;

[0151] The parameter update model for the FedAvg algorithm is as follows:

[0152] ;

[0153] In the formula, , Let represent the global model parameters after the g-th and g+1-th global optimization updates at time t3, respectively. This represents the local training model parameters of the h-th renewable energy power station that can participate in voltage regulation at time t3 after the g+1th global optimization update.

[0154] Based on the above scheme, the accuracy of the trained model is improved by the interaction between the scheduling host and the local terminals of each new energy power station that can participate in voltage regulation. This can improve computational efficiency and alleviate the computational pressure on the scheduling host.

[0155] Based on the above scheme, a multi-scenario judgment criterion for the participation of new energy power plants in distribution network voltage regulation is proposed. This breaks away from the traditional distribution network voltage which only considers exceeding the limit as the sole indicator, increases the reactive power margin and voltage reactive power stability of distribution network substations, and takes into account the coordinated cooperation of reactive power resources at all levels of new energy power plants, so as to efficiently and scientifically guide the coordinated control of new energy and distribution network.

[0156] According to a second aspect of this application, a multi-scenario joint voltage control method for a distribution network is provided, comprising the following steps:

[0157] Collect state space data at the current moment; state space data includes the bus voltage of the substation, the bus voltage of all renewable energy power plants that can participate in voltage regulation, the reactive power of the substation, and the reactive power exchanged between all renewable energy power plants that can participate in voltage regulation and the substation.

[0158] Input the current state space data into the joint voltage control model constructed using the aforementioned method for constructing a multi-scenario joint voltage control model for distribution networks;

[0159] The action space vector at the current moment is output through the joint voltage control model;

[0160] The system collects the second basic data at the current moment, calculates the reactive power of all collectors and all SVGs in the renewable energy power plants that can participate in voltage regulation based on the second basic data at the current moment, and executes the action space vector at the current moment according to the ratio of the reactive power of all collectors and all SVGs. The action space vector includes: the reactive power control quantity of each renewable energy power plant that can participate in voltage regulation, and the reactive power control quantity includes the sum of the reactive power control of the collector inverter and the reactive power control quantity of the SVG.

[0161] Based on the above scheme, after collecting the state space data at the current moment, it can be directly input into the joint voltage control model constructed by the aforementioned method for constructing a multi-scenario joint voltage control model of the distribution network. This will output the action space vector at the current moment. The action space vector includes the reactive power control quantities of each renewable energy power station that can participate in voltage regulation. The reactive power control quantities include the sum of the reactive power control quantities of the collector inverters and the SVG. Then, by collecting the second basic data at the current moment and calculating the reactive power that can be generated by all collectors and all SVGs in the renewable energy power stations that can participate in voltage regulation, the action space vector at the current moment is executed according to the ratio of the reactive power generated by all collectors and all SVGs. This achieves multi-scenario joint voltage control of the distribution network. It avoids internal current flow caused by reactive power imbalance between components and allocates power proportionally according to the adjustable capacity of each collector inverter, SVG, etc. This solves the problems existing in the current whole-station voltage control mode in the prior art.

[0162] In some possible implementations of the second aspect, the second basic data includes the compensable reactive power of the renewable energy power plants that can participate in voltage regulation, the active power, reactive power and rated capacity of all collector lines and collector line inverters, and the capacity, reactive power and set reserved reactive power margin of all SVGs.

[0163] By collecting the second basic data at the current moment, and calculating the reactive power of all collector lines and all SVG in the renewable energy power plants that can participate in voltage regulation, the following data is obtained:

[0164] Calculate the static reactive power reserve factor at the current moment. and the power factor of each renewable energy power station that can participate in voltage regulation ;

[0165] In the formula, This indicates the reactive power of the substation at the current moment. The critical value of reactive power when the voltage of the substation collapses can be obtained from the voltage-reactive power curve of the substation.

[0166] In the formula, This represents the power factor of the i-th renewable energy power station at the current moment. This represents the active power of all collectors in the i-th renewable energy power station at the current moment. This represents the reactive power of all collectors in the i-th renewable energy power station at the current moment;

[0167] Based on the current static reactive power reserve coefficient and the power factor of each renewable energy power station that can participate in voltage regulation, obtain the calculation formula for the current reactive power that can be generated in the current scenario.

[0168] Using the formula for calculating the reactive power that can be generated at the current moment, the reactive power of all collectors and all SVGs in each renewable energy power station that can participate in voltage regulation can be calculated at the current moment.

[0169] The formula for calculating the reactive power available at the current moment is as follows:

[0170] ;

[0171] In the formula, , This represents the reactive power adjustment range of the m-th SVG at the h-th renewable energy power station that can participate in voltage regulation at the current moment. This represents the capacity of the m-th SVG in the h-th renewable energy power station that can participate in voltage regulation. This indicates the reserved reactive power margin of the m-th SVG at the h-th renewable energy power station that can participate in voltage regulation at the current moment. This represents the reactive power of the m-th SVG at the h-th energy station that can participate in voltage regulation at the current moment;

[0172] , This represents the adjustment coefficient of the m-th SVG in the h-th renewable energy power station that can participate in voltage regulation at the current time, where M represents the number of SVGs in the h-th renewable energy power station that can participate in voltage regulation. This indicates the amount of reactive power that the h-th renewable energy power station that can participate in voltage regulation can compensate at the current moment. This value is determined by the reactive power adjustment potential of the renewable energy power station and is provided by each renewable energy power station (not the total upper limit, but the reactive power adjustment range will increase when renewable energy curtails wind and solar power or when active power control is implemented. It is obtained by the renewable energy power station after recalculating the reactive power balance within the station after active power control is implemented). This represents the reactive power generated by the k-th collector line of the h-th renewable energy power station that can participate in voltage regulation at the current time; K represents the number of collector lines in the h-th renewable energy power station that can participate in voltage regulation.

[0173] , These represent the regulation coefficients of the k-th collector line inverter in the h-th renewable energy power station that can participate in voltage regulation at the current time under different scenarios. , This represents the active power of the k-th collector line of the h-th renewable energy power station that can participate in voltage regulation at the current moment. It is collected and calculated by the voltage and current transformers of the power station and then uploaded to the SCADA system, where it can be directly obtained. , This represents the rated capacity of the k-th collector inverter in the h-th renewable energy power station that can participate in voltage regulation.

[0174] Based on the above scheme, using the current static reactive power reserve coefficient and the power factor of each renewable energy power station that can participate in voltage regulation Based on this, the formula for calculating the reactive power that can be generated at the current moment is obtained. By using the formula for calculating the reactive power that can be generated at the current moment, the reactive power of all collectors and all SVGs in each new energy power station at the current moment can be calculated. In other words, the adjustable upper limit value is calculated, and the action space vector is executed according to the proportion of the upper limit value. This adjustment method is more stable and reliable. The reactive power control (AVC) of the new energy power plant is achieved by proportionally distributing reactive power from each SVG (Static Var Generator). The adjustment range is represented by the upper limit of adjustable reactive power, rather than by subtracting the current reactive power transmission value from the maximum adjustable threshold. There are two reasons for this: 1) Proportional distribution of reactive power in the new energy power plant reduces internal circulating currents and increases losses caused by uneven reactive power distribution in the collector inverters and SVGs. If the distribution were based on the remaining capacity, two inverters with different total capacities but the same remaining compensable capacity would be proportionally distributed, resulting in different compensation coefficients for the total capacity of the two collector inverters or SVGs, leading to larger circulating currents and losses. Therefore, proportional distribution based on the total capacity is necessary to determine the maximum adjustable range. 2) When performing reactive power control, the new energy power plant directly issues commands to adjust to a specific value, rather than increasing it by a certain amount. This characteristic description aligns with on-site adjustment practices and avoids further calculations. This solves the problems existing in current voltage control modes, enabling the distribution network to operate safely and stably. The specific methods of both the first and second aspects of this application are executed by the dispatch master station.

[0175] According to a third aspect of this application, a multi-scenario joint voltage control system for a distribution network is provided, including a module for implementing the aforementioned method for constructing a multi-scenario joint voltage control model for a distribution network, or a module for implementing the aforementioned method for multi-scenario joint voltage control of a distribution network.

[0176] In some possible implementations of the third aspect, the control system includes:

[0177] The acquisition module 10 is used to acquire the power characteristic data at time t1;

[0178] The judgment module 20 is used to determine whether the new energy power station needs to participate in voltage control based on power characteristic data;

[0179] Execution module 30 is used to perform model building operations if the new energy power station needs to participate in voltage control; including:

[0180] Acquisition unit 301 is used to acquire historical trend data, which includes basic data and historical state space data.

[0181] Unit 302 is used to select renewable energy power stations that can participate in voltage regulation from all renewable energy power stations based on basic data, and to construct a group of renewable energy power stations.

[0182] The transmitting unit 303 is used to send historical state space data to the new energy power station group for model building and training.

[0183] The receiving unit 304 is used to receive the encrypted voltage control model trained based on historical state space data from each renewable energy power station that can participate in voltage regulation, and to record the number of receptions.

[0184] The decryption unit 305 is used to decrypt the encrypted voltage control model to obtain the decrypted voltage control model;

[0185] The parameter optimization unit 306 is used to optimize the parameters of the decrypted voltage control model to obtain the joint voltage control model.

[0186] In some possible implementations of the third aspect, the judgment module 20 includes:

[0187] The first judgment unit 201 is used to determine whether the static reactive power reserve coefficient of the substation is lower than the coefficient threshold at time t1.

[0188] The second judgment unit 202 is used to determine whether the substation bus voltage at time t1 is not within the preset voltage range if it is not lower than the coefficient threshold, and whether the sum of reactive power absorbed by each new energy station from the substation at time t1 exceeds 50% of the reactive power consumed by the substation at time t1.

[0189] The determination unit 203 is used to determine that if the voltage is below the coefficient threshold, the new energy power station needs to participate in voltage control; it is also used to determine that if the determination condition of the second determination unit 202 is met, the new energy power station needs to participate in voltage control; and it is also used to determine that if the determination condition of the second determination unit 202 is not met, the new energy power station does not need to participate in voltage control.

[0190] In some possible implementations of the third aspect, the selection of unit 302 includes:

[0191] Sensitivity matrix calculation unit 3021 is used to calculate the voltage sensitivity matrix of the new energy power station based on the first basic data;

[0192] Sensitivity matrix processing unit 3022 is used to normalize and binarize the voltage sensitivity matrix to obtain the relationship matrix;

[0193] The reactive power calculation unit 3023 is used to calculate the reactive power of all collectors and all SVGs in each new energy power station based on the second basic data.

[0194] The total electrical feature vector generation unit 3024 is used to construct the electrical feature vector of each new energy power station by combining the reactive power generated by all collectors and all SVGs in each new energy power station with the bus voltage of the corresponding new energy power station; and to merge the electrical feature vectors of each new energy power station to generate the total electrical feature vector.

[0195] Aggregation unit 3025 is used to aggregate node features of the relation matrix and the total electrical feature vector using an aggregation formula to obtain an aggregated vector.

[0196] The node feature matrix construction unit 3026 is used to extract node features from the aggregate vector using a graph attention mechanism and construct the extracted node features into a node feature matrix.

[0197] Transformation unit 3027 is used to perform a linear transformation on the node feature matrix through a fully connected layer to obtain a sample matrix;

[0198] Classification unit 3028 is used to classify the sample matrix through a classification strategy, classifying the new energy power station nodes into: new energy power stations that can participate in voltage regulation and new energy power stations that do not participate in voltage regulation, and constructing the new energy power stations that can participate in voltage regulation into a new energy power station group.

[0199] In some possible implementations of the third aspect, the reactive power calculation unit 3023 includes:

[0200] The first calculation unit 30231 is used to calculate the reactive static reserve coefficient and the power factor of each new energy power station at time t2.

[0201] The scenario and formula acquisition unit 30232 is used to obtain the scenario and the corresponding reactive power calculation formula based on the reactive static reserve coefficient at time t2 and the power factor of each new energy power station.

[0202] The second calculation unit 30233 is used to calculate the reactive power of all collector lines and all SVG in each new energy power station at time t2 using the formula for calculating reactive power.

[0203] In some possible implementations of the third aspect, the control system also includes:

[0204] The first acquisition module 40 is used to acquire the state space data at the current moment; the state space data includes the bus voltage of the substation, the bus voltage of all new energy power plants that can participate in voltage regulation, the reactive power of the substation, and the reactive power exchanged between all new energy power plants that can participate in voltage regulation and the substation.

[0205] The input module 50 is used to input the current state space data into the joint voltage control model constructed by the aforementioned method for constructing a multi-scenario joint voltage control model for a distribution network.

[0206] Output module 60 is used to output the action space vector at the current moment through the joint voltage control model;

[0207] The second acquisition module 70 is used to acquire the second basic data at the current moment;

[0208] The calculation module 80 is used to calculate the reactive power of all collectors and all SVGs in the new energy power plants that can participate in voltage regulation, based on the second basic data at the current moment.

[0209] The voltage control module 90 is used to execute the action space vector at the current moment according to the ratio of the reactive power generated by all collectors to the reactive power generated by all SVGs.

[0210] In some possible implementations of the third aspect, the computation module 80 includes:

[0211] The third calculation unit 801 is used to calculate the current static reactive power reserve coefficient and the power factor of each renewable energy power station that can participate in voltage regulation.

[0212] The current scenario and formula acquisition unit 802 is used to obtain the calculation formula of the current scenario and the corresponding current reactive power based on the current reactive static reserve coefficient and the power factor of each renewable energy power station that can participate in voltage regulation.

[0213] The fourth calculation unit 803 is used to calculate the reactive power of all collectors and all SVGs in each new energy power station that can participate in voltage regulation at the current time, using the calculation formula of the reactive power that can be generated at the current time.

[0214] According to a fourth aspect of this application, an electronic device is provided, comprising:

[0215] Memory;

[0216] Processor; and

[0217] Computer programs;

[0218] The computer program is stored in the memory and configured to be executed by the processor to implement the aforementioned method for constructing a multi-scenario joint voltage control model for a distribution network, or to implement the aforementioned method for joint voltage control of a distribution network in multiple scenarios.

[0219] According to a fifth aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon; the computer program is executed by a processor to implement the aforementioned method for constructing a multi-scenario joint voltage control model for a power distribution network, or is executed to implement the aforementioned method for joint voltage control of a power distribution network in multiple scenarios.

[0220] To make it clearer, the sixth aspect of this application also provides a method for constructing and training an encrypted voltage control model based on historical state-space data for new energy power plants that can participate in voltage regulation, including: a model training method and an encryption method.

[0221] Among some possible implementations of the sixth aspect, the model training methods specifically include:

[0222] The state space of a deep reinforcement learning model is constructed based on the historical state space data of new energy power stations that can participate in voltage regulation.

[0223] Collect local data (including reactive power control quantities of each renewable energy power station that can participate in voltage regulation and active power reduction quantities of renewable energy power stations that can participate in voltage regulation when performing voltage control; reactive power control quantities include the sum of reactive power control quantities of collector line inverters and SVG reactive power control quantities), and construct action space and reward function through local data.

[0224] By training the state space vectors one by one using a deep reinforcement learning model (DQN network model), preliminary voltage control models are obtained for each renewable energy power station that can participate in voltage regulation. Specifically, these models include:

[0225] Initialize the deep reinforcement learning model and its parameters;

[0226] The state-space data of the new energy power station group at time t3 Input the current Q network Training is conducted during this period;

[0227] Select and execute the action space The state-space data of the new energy power station cluster at time t3+1 were obtained. And obtain the reward value through the reward function. ;

[0228] An experience sample set is constructed using the state-space data, action space, and reward value at time t3 and time t3+1. And store the experience sample set in the experience pool;

[0229] At least one set of experience samples is randomly selected from the experience pool and input into the current Q network and the target Q network respectively. The current Q value of the current Q network and the target value of the target Q network are calculated respectively.

[0230] The loss value of the current Q-network is calculated using the loss value calculation formula (local objective function) based on the current Q-value and the target value. The model parameters are then updated based on the loss value using the backpropagation algorithm.

[0231] Update the iteration count;

[0232] Determine if the model has reached the number of iterations T. If it has not reached the number of iterations T, select new state space data and repeat the above steps.

[0233] If the required number of iterations is reached, the trained Q-network is output and used as the initial voltage control model.

[0234] Based on the above scheme, the DQN network model is used to construct and train the initial voltage control model. The model is constructed and trained at the local end of each renewable energy power station that can participate in voltage regulation, which can improve the generalization ability of the constructed model and improve the calculation accuracy.

[0235] In some possible implementations of the sixth aspect, the state space data includes: the bus voltage of the substation, the bus voltage of all renewable energy plants that can participate in voltage regulation, the reactive power of the substation, and the reactive power exchanged between all renewable energy plants that can participate in voltage regulation and the substation.

[0236] The state space is represented as:

[0237] ;

[0238] In the formula, This represents the substation bus voltage at time t3. This represents the bus voltage at time t3 for each renewable energy power station that can participate in voltage regulation. This represents the reactive power of the substation at time t3. This represents the reactive power exchanged between each renewable energy power station that can participate in voltage regulation and the substation at time t3. ;

[0239] The action space is represented as:

[0240] ;

[0241] In the formula, This represents the reactive power control quantity of each renewable energy power station that can participate in voltage regulation at time t3. The reactive power control quantity includes the sum of the reactive power control of the collector-line inverter and the reactive power control quantity of the SVG.

[0242] The reward function is expressed as:

[0243] ;

[0244] In the formula, , , , , , d1 and d2 represent the active power reduction amount when the renewable energy power station that can participate in voltage regulation performs voltage control at time t3, and d3 represents the active power reduction penalty coefficient.

[0245] The formula for calculating the target Q-network is expressed as follows:

[0246] ;

[0247] In the formula, Indicates that the process takes place at time t3. The target value after training the model. This represents the discount factor, which limits the reward value to prevent it from becoming infinitely large; when The reward value obtained by the current Q network when the iteration number T is reached. For the target value; when When the iteration number T has not been reached, the cumulative value is accumulated over a long period of time, meaning the target value includes the immediate reward value and the optimal current Q value calculated in each iteration.

[0248] The formula for calculating the loss value (local objective function) is expressed as follows:

[0249] ;

[0250] In the formula, This indicates that at time t3, after the g-th optimization update (initially g is 0), the h-th renewable energy power station eligible for voltage regulation undergoes [further action / action]. The loss value (local objective function) after model training. This represents the corresponding current Q value. Indicates the number of empirical samples;

[0251] Among some possible implementations of the sixth aspect, the encryption methods specifically include:

[0252] Set the security parameter λ and use the key generation algorithm. Generate public key Private key and homomorphic computation public key ;

[0253] Local training model parameters for renewable energy power plants that can participate in voltage regulation Encryption is performed, specifically including:

[0254] Using public keys Encryption algorithm Generate ciphertext ;

[0255] Set encryption function Using homomorphic computation of public keys Homomorphic operation algorithm Obtain the ciphertext result after homomorphic operation on the encryption function. .

[0256] Correspondingly, the scheduling master station needs to perform decryption, and the decryption method is as follows: using the private key. Decryption algorithm used Decrypting the ciphertext yields the corresponding plaintext, i.e. .

[0257] Based on the above scheme, homomorphic encryption is used to transmit encrypted data between the scheduling master station and the new energy power station group. The encryption and decryption process is simple and highly secure, which can improve data processing efficiency, increase data privacy protection, and reduce communication costs.

[0258] It should be understood that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order constraint on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the diagram may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0259] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as C, VHDL, Verilog, the object-oriented programming language Java, and the interpreted scripting language JavaScript.

[0260] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0261] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0262] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0263] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0264] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0265] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for constructing a multi-scenario joint voltage control model for a power distribution network, characterized in that, include: Obtain the power characteristic data at time t1, and determine whether the new energy power station needs to participate in voltage control based on the power characteristic data; If the renewable energy power station needs to participate in voltage control, then perform the following operations: Acquire historical trend data, which includes basic data and historical state space data; Based on basic data, select renewable energy power stations that can participate in voltage regulation from all renewable energy power stations and construct a group of renewable energy power stations; Historical state space data is sent to a group of new energy power stations for model building and training; Receive encrypted voltage control models trained based on historical state space data from each renewable energy power station that can participate in voltage regulation, and record the number of times the models are received. The encrypted voltage control model is decrypted to obtain the decrypted voltage control model. The parameters of the decrypted voltage control model are optimized to obtain a joint voltage control model.

2. The method for constructing a multi-scenario joint voltage control model for a distribution network according to claim 1, characterized in that, The power characteristic data includes the substation bus voltage, the reactive power absorbed by each new energy power plant from the substation, and the substation's reactive power static reserve coefficient. The determination of whether renewable energy power plants need to participate in voltage control based on power characteristic data specifically includes: Determine whether the static reactive power reserve coefficient of the substation is lower than the coefficient threshold at time t1; If the voltage is below the coefficient threshold, it is directly determined that the new energy power station needs to participate in voltage control. If it is not lower than the coefficient threshold, then determine whether the substation bus voltage at time t1 is not within the preset voltage range, and whether the sum of reactive power absorbed by each new energy power station from the substation at time t1 exceeds 50% of the reactive power consumed by the substation at time t1; If so, it is determined that the new energy power station needs to participate in voltage control; Otherwise, it is determined that the participation of new energy power stations in voltage control is not required.

3. The method for constructing a multi-scenario joint voltage control model for a distribution network according to claim 1, characterized in that, The basic data includes first basic data and second basic data; The process of selecting renewable energy power stations that can participate in voltage regulation from all renewable energy power stations based on basic data and constructing them into a renewable energy power station group specifically includes: The voltage sensitivity matrix of the new energy power station is calculated based on the first basic data. The voltage sensitivity matrix is ​​then normalized and binarized to obtain the relationship matrix. Based on the second basic data, calculate the reactive power of all collectors and all SVG in each new energy power station, and construct the electrical characteristic vector of each new energy power station together with the bus voltage of the corresponding new energy power station; merge the electrical characteristic vectors of each new energy power station to generate the total electrical characteristic vector. The relationship matrix and the total electrical feature vector are aggregated using an aggregation formula to obtain an aggregated vector. A graph attention mechanism is then used to extract the node features from the aggregated vector, and the extracted node features are used to construct a node feature matrix. The node feature matrix is ​​linearly transformed by a fully connected layer to obtain a sample matrix. The sample matrix is ​​then classified using a classification strategy, and the new energy power station nodes are classified into two categories: new energy power stations that can participate in voltage regulation and new energy power stations that do not participate in voltage regulation. The new energy power stations that can participate in voltage regulation are then constructed into a new energy power station group.

4. The method for constructing a multi-scenario joint voltage control model for a distribution network according to claim 1, characterized in that, The parameter optimization of the decryption voltage control model to obtain the joint voltage control model specifically includes: The model parameters in the decrypted voltage control model are globally optimized and updated using the FedAvg algorithm. Determine if the number of receptions has reached the preset number. If not, send the globally optimized and updated model parameters to the new energy power station group for model retraining. If the preset number of iterations is reached, the globally optimized and updated model parameters are output, resulting in a joint voltage control model.

5. The method for constructing a multi-scenario joint voltage control model for a distribution network according to claim 3, characterized in that, The calculation of the reactive power generated by all collectors and all SVG in each new energy power station based on the second basic data specifically includes: Calculate the static reactive power reserve factor and the power factor of each renewable energy power station at time t2; Based on the static reactive power reserve coefficient at time t2 and the power factor of each new energy power station, obtain the calculation formula for the scenario and the corresponding reactive power that can be generated. The reactive power of all collectors and all SVGs in each new energy power station at time t2 is calculated using the formula for calculating reactive power.

6. A method for joint voltage control in multiple scenarios of a power distribution network, characterized in that, Includes the following steps: Collect state-space data at the current moment; The current state space data is input into the joint voltage control model constructed by the method described in any one of claims 1 to 5 for constructing a multi-scenario joint voltage control model for a distribution network. The action space vector at the current moment is output through the joint voltage control model; Collect the second basic data at the current moment, calculate the reactive power of all collectors and all SVGs in the new energy power plants that can participate in voltage regulation based on the second basic data at the current moment, and execute the action space vector at the current moment according to the ratio of the reactive power of all collectors and the reactive power of all SVGs.

7. The multi-scenario joint voltage control method for distribution networks according to claim 6, characterized in that, The calculation of the reactive power generated by all collectors and all SVGs in the renewable energy power plants eligible for voltage regulation, based on the second basic data at the current moment, specifically includes: Calculate the current static reactive power reserve coefficient and the power factor of each renewable energy power station that can participate in voltage regulation; Based on the current static reactive power reserve coefficient and the power factor of each renewable energy power station that can participate in voltage regulation, obtain the calculation formula for the current reactive power that can be generated in the current scenario. Using the formula for calculating the reactive power that can be generated at the current moment, the reactive power of all collectors and all SVGs in each renewable energy power station that can participate in voltage regulation can be calculated at the current moment.

8. A multi-scenario joint voltage control system for power distribution networks, characterized in that, It includes a module that implements the construction method of the distribution network multi-scenario joint voltage control model as described in any one of claims 1 to 5, or includes a module that implements the distribution network multi-scenario joint voltage control method as described in claim 6 or 7.

9. An electronic device, characterized in that, include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method for constructing a multi-scenario joint voltage control model for a distribution network as described in any one of claims 1 to 5, or to implement the multi-scenario joint voltage control method for a distribution network as described in claim 6 or 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program; the computer program is executed by a processor to implement the method for constructing a multi-scenario joint voltage control model for a distribution network as described in any one of claims 1 to 5, or to implement the multi-scenario joint voltage control method for a distribution network as described in claim 6 or 7.

Images