A physical-data fusion-based double-time-scale active power distribution network voltage optimization control method

Abstract

The application discloses a kind of physical-data fusion under double time scale active distribution network voltage optimization control method, belong to electric power technical field.For solving the problem that existing single time scale regulation ability is limited, data-driven method lacks physical security guarantee and fast and slow regulation equipment is insufficient in cooperation, the application constructs double time scale operating state, and slow time scale state is introduced into fast time scale to form extended state;In slow time scale, structural regulation instruction is generated based on physical model and voltage safety operation boundary is determined;In fast time scale, candidate regulation instruction is generated by data generation based on extended state;The safety of candidate instruction is predicted and corrected online through physical model, and the final regulation instruction meeting the physical safety constraint is generated;After executing final instruction, update operating state and feedback to fast and slow scale, form closed-loop control.The application realizes the safe, stable and efficient collaborative control of active distribution network voltage.

Description

Technical Field

[0001] This invention belongs to the field of power technology, and in particular relates to a method for voltage optimization control of active distribution networks with dual time scales under physical-data fusion. Background Technology

[0002] With the widespread integration of distributed power sources, electric vehicles, and power electronic equipment into distribution networks, these networks exhibit multi-source access and multi-point regulation characteristics. The rapid fluctuations in renewable energy output and load over time result in strong uncertainty and time-series correlation in the voltage operation of distribution networks, posing new challenges to traditional voltage control technologies. Existing methods, such as voltage control based on physical models, regulate distribution network voltage by adjusting transformer taps and capacitor banks. While these methods have clear physical meaning and high safety, their adjustment cycles are relatively long, making it difficult to respond promptly to rapid changes in renewable energy output and load. Furthermore, these methods are highly dependent on system parameters and operating models, limiting their effectiveness under complex operating conditions. Intelligent control methods based on operational data can characterize the voltage operation characteristics of distribution networks using historical data and real-time status information, offering advantages in handling nonlinearity and uncertainty. However, their decision-making process lacks explicit characterization of the physical constraints of the distribution network system, posing a risk of generating unsafe voltage control commands and making them difficult to directly apply to practical distribution network engineering. Therefore, existing power distribution network voltage control methods suffer from limited single-time-scale regulation capabilities, a lack of physical security guarantees in data-driven methods, and insufficient coordination between fast and slow regulation devices. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a dual-timescale active distribution network voltage optimization control method under physical-data fusion, thereby resolving the issues present in the prior art.

[0004] Firstly, to achieve the above objectives, this invention provides a method for optimizing and controlling the voltage of an active distribution network under physical-data fusion with dual time scales, comprising the following steps: S1. Construct the operating state of the distribution network under dual time scales, wherein the fast time scale state is constructed based on real-time and historical observation data, and the slow time scale state is constructed based on statistical characteristic quantities within a slow cycle, and the slow time scale state is introduced into the fast time scale state set to form an extended state. S2. On a slow time scale, solve the optimization problem based on the physical model, generate the regulation command of the structural regulation device, and determine the voltage safety operation boundary for constraining the fast time scale control. S3. On a fast timescale, using the extended state as input, generate candidate adjustment commands for the fast adjustment device through a data-driven model; S4. Before execution, the candidate regulation command is predicted for safety based on the physical model. If the prediction result violates the voltage safety operation boundary, it is corrected online to generate the final regulation command that meets the physical safety constraints. S5. Issue the final adjustment command to update the distribution network operating status, and feed back the execution results to the fast time scale and slow time scale status updates respectively to form a closed-loop control.

[0005] Optionally, the process of constructing the operating state of the distribution network under dual time scales in S1 includes: in the fast time scale, collecting node voltage, active power, reactive power and auxiliary variables to form an observation vector and forming a historical observation sequence; in the slow time scale, statistically analyzing the voltage and power within a slow cycle to construct a slow-scale state vector; and through a state fusion mapping function, fusing the slow-scale state vector with the historical observation sequence to generate an extended state for fast-scale regulation.

[0006] Optionally, the process of generating regulation commands for structural regulation equipment in S2 includes: constructing a voltage prediction model based on the distribution network physical model, using the slow-scale state vector as input, and predicting the node voltage after the structural regulation command is applied; with the goal of minimizing voltage deviation and minimizing the regulation change of slow-speed equipment, solving the optimization problem under the constraints of the equipment itself, and obtaining the structural regulation command composed of continuous and discrete variables.

[0007] Optionally, the process of determining the voltage safety operating boundary in S2 includes: obtaining the node voltage prediction result in the slow cycle according to the structural adjustment command and the voltage prediction model; and generating the voltage safety operating boundary for constraining the subsequent fast time scale control process based on the node voltage prediction result and the preset node voltage allowable upper and lower bounds.

[0008] Optionally, the process of generating candidate regulation instructions for fast-regulating devices in S3 includes: inputting the extended state into a data-driven decision mapping function, the function outputting an original regulation decision consisting of active power regulation and reactive power regulation; determining the voltage regulation demand under a fast time scale based on the deviation between the node voltage and the reference voltage at the current moment, the original regulation decision being the candidate regulation instruction.

[0009] Optionally, the process of making a safety prediction of the candidate regulation command in S4 includes: based on the distribution network physical model, using the voltage sensitivity matrix of active and reactive power regulation, calculating the node voltage change caused by the action of the candidate regulation command; and superimposing the current node voltage with the voltage change to obtain the voltage prediction value after the execution of the candidate command.

[0010] Optionally, the process of online correction of candidate regulation commands in S4 includes: if the voltage prediction value violates the voltage safety operation boundary, then with the goal of minimizing the difference between the corrected regulation command and the candidate regulation command, and with the voltage safety operation boundary and the device's own regulation capability as constraints, an online correction model is constructed; the online correction model is solved to obtain the final regulation command that satisfies the physical safety constraints.

[0011] Optionally, the process of updating the execution result feedback in S5 includes: using the system operating state after the final adjustment command is executed to update the observation vector and historical observation sequence of the next fast time scale; at the end of a slow cycle, summarizing the fast-scale execution results within that cycle to form a slow-scale feedback quantity containing voltage statistical characteristics or safety correction frequency, and updating the slow-scale state vector of the next slow cycle accordingly.

[0012] Secondly, the present invention also provides a computer terminal device, comprising: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the physical-data fusion dual-timescale active distribution network voltage optimization control method in the first aspect described above.

[0013] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the steps of the physical-data fusion-based dual-timescale active distribution network voltage optimization control method described in the first aspect above.

[0014] Compared with the prior art, the present invention has the following advantages and technical effects: This invention provides a dual-timescale voltage optimization control method for active distribution networks under physical-data fusion. By constructing a control framework that integrates the physical model of the distribution network voltage with operational data, physical constraints are introduced into the voltage control execution process, taking into account the impact of physical constraints on the safety and reliability of voltage regulation. A dual-timescale collaborative control approach is adopted, controlling the response characteristics of different regulating devices separately, effectively avoiding the problems of regulation conflicts and frequent device actions found in single-timescale control. At the fast timescale, operational data is used to rapidly suppress voltage fluctuations, while at the slow timescale, structural adjustments are used to maintain the overall system operating state, improving the voltage stability of the distribution network under conditions of high renewable energy penetration. This invention does not rely on specific algorithms or model structures, possesses good versatility, scalability, and portability, and is applicable to active distribution networks of different scales and operating conditions. The implementation process is clear, with strong engineering feasibility, and has significant engineering application value. Attached Figure Description

[0015] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart illustrating a physical-data fusion-based active distribution network voltage optimization control method according to an embodiment of the present invention. Detailed Implementation

[0016] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0017] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0018] Example 1 like Figure 1 As shown, this embodiment provides a method for voltage optimization control of active distribution networks with dual time scales under physical-data fusion, including: S1. Construct the operating state of the distribution network under dual time scales, wherein the fast time scale state is constructed based on real-time and historical observation data, and the slow time scale state is constructed based on statistical characteristic quantities within a slow cycle, and the slow time scale state is introduced into the fast time scale state set to form an extended state. S2. On a slow time scale, solve the optimization problem based on the physical model, generate the regulation command of the structural regulation device, and determine the voltage safety operation boundary for constraining the fast time scale control. S3. On a fast timescale, using the extended state as input, generate candidate adjustment commands for the fast adjustment device through a data-driven model; S4. Before execution, the candidate regulation command is predicted for safety based on the physical model. If the prediction result violates the voltage safety operation boundary, it is corrected online to generate the final regulation command that meets the physical safety constraints. S5. Issue the final adjustment command to update the distribution network operating status, and feed back the execution results to the fast time scale and slow time scale status updates respectively to form a closed-loop control.

[0019] Furthermore, the process of constructing the operating state of the distribution network under dual time scales in S1 includes: in the fast time scale, collecting node voltage, active power, reactive power and auxiliary variables to form an observation vector and forming a historical observation sequence; in the slow time scale, statistically analyzing the voltage and power within a slow cycle to construct a slow-scale state vector; and through a state fusion mapping function, fusing the slow-scale state vector with the historical observation sequence to generate an extended state for fast-scale regulation.

[0020] Specifically, the implementation process of this embodiment includes: Step 1: Construction of dual-time-scale distribution network voltage operation status and preparation of control inputs: This step is used to construct the system state information required for dual-timescale voltage optimization control of the distribution network, providing a unified input for slow-timescale structural regulation and fast-timescale real-time distribution network voltage control.

[0021] Construction of fast timescale observations: At a fast timescale t The observation vector of the distribution network system is defined as: o t =[U t , P t Q t , z t ]; in, U t Indicates time t The node voltage magnitude vector; P t and Q t These represent the active and reactive power injection vectors at the nodes, respectively; z t These represent auxiliary variables related to the operating status of the distribution network, including but not limited to transformer tap positions, reactive power compensation device switching status, and renewable energy output levels.

[0022] To construct a fast timescale input with temporal information, a length of T is defined. f The historical observation sequence is as follows: ={o t-Tf+1 ,o t-Tf+2 ,…,o t}; in, Indicates the first t The historical observation state sequence corresponding to a fast time scale is used to describe the changing trend of variables such as voltage and power in the distribution network in a short period of time, providing time series data for subsequent data-driven fine adjustment.

[0023] Slow-timescale state construction: periodic on a slow time scalek Internally, to describe the slower changing trends and structural background of the system, a slow-timescale state vector is constructed: ; in, These represent the statistical characteristics of voltage, active power, and reactive power within the slow time scale period of the distribution network, respectively, and can be expressed as average value, extreme value, or fluctuation range. This represents state variables related to the operation of slow-speed regulating equipment, such as tap position and switching frequency statistics.

[0024] Fast timescale extended state construction: To ensure information consistency between fast and slow timescale controls, when constructing the fast timescale state, the current slow timescale state is introduced as constraint information into the fast timescale state set.

[0025] Define the extended state for fast timescale adjustment as follows: ; in, This is a state fusion mapping function, whose input is a fast-timescale historical state sequence. With slow timescale state s k The output is an extended state vector for fast time-scale adjustment. .

[0026] To improve the numerical stability of subsequent decision-making processes, the state vector can be normalized. For any vector to be normalized... x (can be o) t s k or The components of the equation are defined as follows: ; in, X This represents the state vector to be normalized, which can be a fast-timescale extended state or a slow-timescale state; X min and X max These are the physical lower and upper limits of the corresponding state variables, respectively; the normalized state serves as the unified input for the data-driven decision module and the physical model channel in subsequent steps.

[0027] Furthermore, the process of generating regulation commands for structural regulation equipment in S2 includes: constructing a voltage prediction model based on the distribution network physical model, using the slow-scale state vector as input, and predicting the node voltage after the structural regulation command is applied; with the goal of minimizing voltage deviation and minimizing the regulation change of slow-speed equipment, solving the optimization problem under the constraints of the equipment itself, and obtaining the structural regulation command composed of continuous and discrete variables.

[0028] Furthermore, the process of determining the voltage safety operation boundary in S2 includes: obtaining the node voltage prediction result in the slow cycle according to the structural adjustment command and the voltage prediction model; and generating the voltage safety operation boundary for constraining the subsequent fast time scale control process based on the node voltage prediction result and the preset node voltage allowable upper and lower bounds.

[0029] Specifically, the implementation process of this embodiment includes: Step 2: Slow-timescale structural regulation and voltage security constraint generation: This step performs structural voltage regulation on the distribution network operation status at a slow time scale. It generates slow time scale regulation decisions through physical models and optimization methods, and further determines the distribution network voltage safety operation boundary to constrain real-time control at a fast time scale.

[0030] Slow timescale regulation does not directly act on every fast timescale voltage control moment, but adjusts the overall operating structure of the system at a lower update frequency, providing a stable physical operating space and safety constraints for fast timescale control.

[0031] periodic on a slow time scale k Within this framework, the slow-time-scale structural adjustment decision vector is defined as follows: ; Where, α k This represents a continuous structural adjustment variable at a slow time scale, used to describe the adjustment range of slow-adjustable equipment; β k =[ β 1,k ,β 2,k, ..., β M,k ] represents a discrete structure adjustment variable vector, used to describe the operating status of slow-switching or gear-type equipment; β m,k ∈{0, 1} represents m structural adjustment devices in a slow timescale period. k The internal operating status; M represents the number of structural adjustment devices on a slow time scale.

[0032] Based on the distribution network topology, line parameters, and operating status, a voltage prediction model under a slow time scale is constructed to characterize the impact of slow time scale regulation decisions on node voltage levels. The node voltage prediction vector under the slow time scale is defined as follows: ; Where g(·) represents the mapping function based on the physical model of the electrical system, which is obtained from the power flow calculation model or its linear / approximate form; This indicates that structural adjustment decisions are made within a slow timescale period k. The node voltage prediction results obtained after the action.

[0033] The objective of slow-time-scale structural regulation is to minimize voltage deviation in the distribution network system while satisfying safety constraints and limiting the operating frequency of slow-moving equipment, and to reserve a margin for fast-time-scale regulation. To obtain reasonable structural regulation decisions under slow-time-scale conditions, the following optimization objective function is constructed: ; in, U ref Represents the node voltage reference value vector; a k-1 and β k-1 These represent the continuous and discrete structural adjustment variables that have been executed in the previous slow timescale period, respectively. These are the voltage deviation weighting coefficient, the continuous structure regulation smoothing weighting coefficient, and the discrete structure regulation switching penalty weighting coefficient, respectively, used to balance voltage deviation suppression and regulation smoothness.

[0034] The above optimization objective is solved under the following constraints: ; ; In obtaining structural adjustment decisions on slow timescales Then, based on the voltage prediction results To further determine the safe operating boundary of distribution network voltage within a slow time scale period: ; in, U min and U max These represent the lower and upper bounds of the allowable voltage at the distribution network nodes, respectively, and are used to constrain the voltage variation range at each moment during the fast time-scale control process in step 3.

[0035] Furthermore, the process of generating candidate regulation commands for fast-regulating devices in S3 includes: inputting the extended state into a data-driven decision mapping function, the function outputting an original regulation decision consisting of active power regulation and reactive power regulation; determining the voltage regulation demand on a fast time scale based on the deviation between the node voltage and the reference voltage at the current moment, the original regulation decision being the candidate regulation command.

[0036] Specifically, the implementation process of this embodiment includes: Step 3: Generation of candidate adjustment instructions driven by fast timescale data: In this step, under the constraints of the slow-time-scale structural regulation and voltage safety operation boundary in step 2, candidate regulation commands for real-time regulation are generated based on the fast-time-scale extended state. The generated candidate regulation commands are used to reflect the regulation of the current operating state by the data-driven model. They themselves have not yet guaranteed to meet the physical safety constraints. Whether they are executable will be further judged and corrected online in step 4.

[0037] Definition of the first t The adjustment decision vector at each fast time scale is: ; in, Indicates L fast adjustment devices at time... t The active power regulation vector; This represents the reactive power regulation vector of the corresponding device; L is the number of controllable devices on a fast time scale.

[0038] Based on extended state Data-driven decision mapping generates initial adjustment decisions on a fast timescale: ; in, πf (·) is a fast timescale data-driven decision mapping function used to describe the nonlinear relationship from state to adjustment decision; This is the original adjustment decision that does not take into account physical security constraints.

[0039] To characterize the voltage regulation demand of the distribution network on a fast time scale, the node voltage deviation vector is defined as: ; in, U t Representing a fast timescale t The node voltage vector; U ref This represents a vector of node voltage reference values. To ensure that regulation commands can be directly executed by the equipment, the initial regulation decisions must satisfy the equipment's own constraints. Voltage deviation. ΔU t This is used to characterize the difference between the current operating state of the system and the expected voltage level, providing a basis for safety judgment and online correction in the subsequent step 4.

[0040] Furthermore, the process of making a safety prediction of the candidate regulation command in S4 includes: based on the distribution network physical model, using the voltage sensitivity matrix of active and reactive power regulation, calculating the node voltage change caused by the action of the candidate regulation command; and superimposing the current node voltage with the voltage change to obtain the voltage prediction value after the execution of the candidate command.

[0041] Furthermore, the online correction process for candidate regulation commands in S4 includes: if the voltage prediction value violates the voltage safety operating boundary, then with the goal of minimizing the difference between the corrected regulation command and the candidate regulation command, and with the voltage safety operating boundary and the device's own regulation capability as constraints, an online correction model is constructed; the online correction model is solved to obtain the final regulation command that satisfies the physical safety constraints.

[0042] Specifically, the implementation process of this embodiment includes: Step 4: Physical-data coordinated control and online voltage correction of distribution networks across time scales: Before the fast-timescale adjustment command is executed, this step introduces a physical model of the distribution network system to determine the safety of the candidate adjustment commands generated in step 3, and corrects them online if they do not meet the safety constraints, thereby generating the final executable adjustment command that meets the physical safety constraints.

[0043] By performing physical safety assessments and online corrections before execution, the feasibility and security of adjustment commands in actual power distribution systems can be ensured without altering the adjustment intent of the data-driven model.

[0044] To determine the voltage safety of candidate regulation commands after execution, the voltage changes caused by the candidate regulation commands are predicted based on the physical model of the electrical system. This prediction is made at fast timescales. t The node voltage change prediction vector under the action of candidate regulation commands is defined as follows: ; in, and These represent the active and reactive power adjustment vectors in the candidate adjustment commands, respectively. This is the sensitivity matrix of node voltage to active power regulation on a fast time scale; Sensitivity matrix of node voltage to reactive power regulation on fast time scale; To predict voltage changes; N is the number of nodes, and L is the number of controllable resources participating in fast timescale regulation.

[0045] This yields the predicted voltage values ​​for the distribution network on a fast timescale: ; Based on the above voltage prediction results, it is determined whether the candidate regulation commands meet the voltage safety operation boundary generated by the slow time scale: ; When a candidate regulation command meets the above voltage safety constraints, the regulation command is physically feasible under the current operating state and is directly executed as a safety regulation decision. ; When the original regulation decision does not meet the distribution network voltage security constraints, an online distribution network voltage security correction model is constructed to perform a minimum-cost correction of the original regulation decision while satisfying physical and equipment constraints. The fast-timescale distribution network voltage security regulation decision is defined as follows: ; ; in, This is the adjustment command vector to be corrected.

[0046] By solving the above optimization problem, a safety adjustment command that satisfies physical safety constraints is obtained. .

[0047] Furthermore, the process of updating the execution results in S5 includes: using the system operating state after the final adjustment command is executed to update the observation vector and historical observation sequence of the next fast time scale; at the end of a slow cycle, summarizing the fast-scale execution results within that cycle to form a slow-scale feedback quantity containing voltage statistical characteristics or safety correction frequency, and updating the slow-scale state vector of the next slow cycle accordingly.

[0048] Specifically, the implementation process of this embodiment includes: Step 5: Execution of safety regulation commands and cross-timescale distribution network voltage feedback update: This step applies the voltage regulation command that satisfies physical safety constraints generated in step 4 to the actual distribution network system, updates the voltage operating status of the distribution network system, and feeds back the execution results to the distribution network voltage control process on both the fast and slow time scales. This realizes the execution, dynamic evolution, and cross-time scale voltage information transmission of the distribution network voltage regulation command, forming a dual-time scale collaborative closed-loop control of the distribution network voltage.

[0049] At a fast timescale t Safety adjustment instructions It operates on the corresponding controllable resources in the distribution network, performing corresponding active and reactive voltage regulation operations. After the regulation command is executed, based on the power flow calculation model or equivalent physical model of the distribution network, the system operating state is updated to obtain the node voltage, power distribution, and other operational measurement data at the next fast time scale.

[0050] Based on the system operation results obtained after executing the safety adjustment command in this step, update the fast timescale observation vector. t+1 And update the historical observation sequence on a fast timescale accordingly: ; The updated fast timescale historical observation sequence will be used as input for the fast timescale extended state construction in step 1, and will be used to generate candidate adjustment instructions for the next fast timescale time t+1.

[0051] periodic on a slow time scale k Within this system, as fast-timescale distribution network voltage control is continuously executed, the system operation results are statistically analyzed and summarized to form slow-timescale feedback quantities for distribution network voltage. These slow-timescale feedback quantities may include, but are not limited to, the following: statistical characteristics of node voltages within the slow-timescale period; the average execution magnitude of distribution network voltage regulation commands; and operational indicators such as the number of times distribution network voltage exceeds limits or the frequency of safety corrections.

[0052] When the slow timescale update condition is met, the slow timescale state vector is updated based on the above statistical feedback information: ; in, Ψ ( · ) represents the slow-time-scale state update mapping function.

[0053] Updated slow-timescale state s k+1 This will serve as the input for the structural adjustment in step 2 of the next slower timescale cycle.

[0054] At the fast timescale level, if the current fast timescale time t does not meet the slow timescale period termination condition, proceed to the next fast timescale time t+1 and repeat steps 3 to 5.

[0055] When the termination condition of a slow timescale period is met, the current slow timescale period k is completed, and the next slow timescale period begins. k +1, and repeat steps 2 through 5.

[0056] By executing distribution network voltage regulation commands, updating system operating status, and performing cross-timescale feedback, this step achieves an information closed loop between fast timescale control and slow timescale regulation, enabling the physical-data fusion dual timescale control method to operate continuously and stably in the distribution network system.

[0057] Example 2 In this embodiment, a computer terminal device is provided, including: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-described physical-data fusion dual-timescale active distribution network voltage optimization control method.

[0058] In this embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-described physical-data fusion-based dual-timescale active distribution network voltage optimization control method.

[0059] This invention provides a dual-timescale voltage optimization control method for active distribution networks under physical-data fusion. By constructing a control framework that integrates the physical model of the distribution network voltage with operational data, physical constraints are introduced into the voltage control execution process, taking into account the impact of physical constraints on the safety and reliability of voltage regulation. A dual-timescale collaborative control approach is adopted, controlling the response characteristics of different regulating devices separately, effectively avoiding the problems of regulation conflicts and frequent device actions found in single-timescale control. At the fast timescale, operational data is used to rapidly suppress voltage fluctuations, while at the slow timescale, structural adjustments are used to maintain the overall system operating state, improving the voltage stability of the distribution network under conditions of high renewable energy penetration. This invention does not rely on specific algorithms or model structures, possesses good versatility, scalability, and portability, and is applicable to active distribution networks of different scales and operating conditions. The implementation process is clear, with strong engineering feasibility, and has significant engineering application value.

[0060] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for voltage optimization control of active distribution networks with dual time scales under physical-data fusion, characterized in that, Includes the following steps: S1. Construct the operating state of the distribution network under dual time scales, wherein the fast time scale state is constructed based on real-time and historical observation data, and the slow time scale state is constructed based on statistical characteristic quantities within a slow cycle, and the slow time scale state is introduced into the fast time scale state set to form an extended state. S2. On a slow time scale, solve the optimization problem based on the physical model, generate the regulation command of the structural regulation device, and determine the voltage safety operation boundary for constraining the fast time scale control. S3. On a fast timescale, using the extended state as input, generate candidate adjustment commands for the fast adjustment device through a data-driven model; S4. Before execution, the candidate regulation command is predicted for safety based on the physical model. If the prediction result violates the voltage safety operation boundary, it is corrected online to generate the final regulation command that meets the physical safety constraints. S5. Issue the final adjustment command to update the distribution network operating status, and feed back the execution results to the fast time scale and slow time scale status updates respectively to form a closed-loop control.

2. The method according to claim 1, characterized in that, The process of constructing the operating state of the distribution network under dual time scales in S1 includes: on the fast time scale, collecting node voltage, active power, reactive power and auxiliary variables to form an observation vector and forming a historical observation sequence; on the slow time scale, statistically analyzing the voltage and power within a slow cycle to construct a slow-scale state vector; and through a state fusion mapping function, fusing the slow-scale state vector with the historical observation sequence to generate an extended state for fast-scale regulation.

3. The method according to claim 1, characterized in that, The process of generating regulation commands for structural regulation equipment in S2 includes: constructing a voltage prediction model based on the distribution network physical model, using the slow-scale state vector as input, and predicting the node voltage after the structural regulation command is applied; with the goal of minimizing voltage deviation and minimizing the regulation change of slow-speed equipment, solving the optimization problem under the constraints of the equipment itself, and obtaining the structural regulation command composed of continuous and discrete variables.

4. The method according to claim 3, characterized in that, The process of determining the voltage safety operation boundary in S2 includes: obtaining the node voltage prediction result in the slow cycle according to the structural adjustment command and the voltage prediction model; and generating the voltage safety operation boundary for constraining the subsequent fast time scale control process based on the node voltage prediction result and the preset node voltage allowable upper and lower bounds.

5. The method according to claim 1, characterized in that, The process of generating candidate regulation commands for fast-regulating devices in S3 includes: inputting the extended state into a data-driven decision mapping function, the function outputting an original regulation decision consisting of active power regulation and reactive power regulation; determining the voltage regulation demand under a fast time scale based on the deviation between the node voltage and the reference voltage at the current moment, the original regulation decision being the candidate regulation command.

6. The method according to claim 1, characterized in that, The process of making a safety prediction of the candidate regulation command in S4 includes: based on the distribution network physical model, using the voltage sensitivity matrix to active and reactive power regulation, calculating the node voltage change caused by the action of the candidate regulation command; and superimposing the current node voltage with the voltage change to obtain the voltage prediction value after the execution of the candidate command.

7. The method according to claim 1, characterized in that, The process of online correction of candidate regulation commands in S4 includes: if the voltage prediction value violates the voltage safety operation boundary, then with the goal of minimizing the difference between the corrected regulation command and the candidate regulation command, and with the voltage safety operation boundary and the device's own regulation capability as constraints, an online correction model is constructed; the online correction model is solved to obtain the final regulation command that satisfies the physical safety constraints.

8. The method according to claim 1, characterized in that, The process of updating the execution results in S5 includes: using the system operating state after the final adjustment command is executed to update the observation vector and historical observation sequence of the next fast time scale; at the end of a slow cycle, summarizing the fast-scale execution results within that cycle to form a slow-scale feedback quantity containing voltage statistical characteristics or safety correction frequency, and updating the slow-scale state vector of the next slow cycle accordingly.

9. A computer terminal device, characterized in that, include: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the steps of the method as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-8.

Images