Power distribution network reactive power distribution acceleration method and system based on industrial-grade configuration platform

Abstract

The invention relates to the technical field of power grid acceleration, and discloses a power distribution network reactive power distribution acceleration method and system based on an industrial-grade configuration platform. The method comprises the following steps of: constructing a reactive power distribution optimization model: establishing the reactive power distribution optimization model by taking minimization of the total reactive power regulation cost of the transformer area as an objective function and taking meeting of the reactive power demand of the transformer area as a constraint condition; the reactive power demand of the transformer area is modeled into bounded uncertain parameters changing in a determined interval, the two-stage robust optimization decision framework is adopted, the first stage decides the switching state of a discrete equipment capacitor, and the second stage decides the switching state of the discrete equipment capacitor after an uncertain scene is revealed. Reactive power output of the static reactive power compensator is continuously adjusted; solving a model and generating a strategy; operation acceleration is realized based on an industrial-grade configuration platform. The improvement of the method is that operation acceleration of the reactive power distribution model of the power distribution network is realized based on a multi-configuration parallel computing architecture of an industrial-grade configuration platform.

Description

Technical Field

[0001] This invention belongs to the field of power grid acceleration technology, specifically relating to a method and system for accelerating reactive power distribution in distribution networks based on an industrial-grade configuration platform. Background Technology

[0002] The large-scale integration of distributed photovoltaic and other renewable energy sources into power distribution areas, coupled with the widespread adoption of electric vehicles leading to a sharp increase in the proportion of charging pile load in residential electricity consumption, has resulted in highly volatile reactive power demand in these areas. This poses a significant safety hazard, potentially causing transformer overload and voltage exceeding limits. Traditional reactive power allocation models (such as deterministic optimization methods) suffer from high computational complexity and slow solution speed when dealing with such uncertainties, making it difficult to meet real-time scheduling requirements. Upgrading and transforming distribution network equipment is time-consuming and costly, and existing optimization algorithms often fail to achieve sub-second response times when dealing with high-proportion renewable energy fluctuations due to limited computational resources, further exacerbating system operational risks.

[0003] Industrial-grade configuration platforms, through their multi-configuration parallel computing architecture, can effectively address the aforementioned bottlenecks. However, in existing technologies, the application of such platforms in the field of reactive power allocation in distribution networks is insufficient, especially when dealing with robust optimization models (such as two-stage decision frameworks). The lack of efficient parallel acceleration mechanisms leads to excessively long iterative solution processes, such as column constraint generation algorithms, which cannot dynamically adapt to real-time changes in reactive power demand. Therefore, there is an urgent need to develop reactive power allocation acceleration methods based on industrial-grade configuration platforms to improve computational efficiency, ensure system robustness, and provide technical support for building new smart distribution networks. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for accelerating reactive power distribution in power distribution networks based on an industrial-grade configuration platform, which can effectively solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: a method for accelerating reactive power distribution in power distribution networks based on an industrial-grade configuration platform, comprising the following specific steps: Constructing a reactive power allocation optimization model: With minimizing the total cost of reactive power regulation in the transformer area as the objective function and meeting the reactive power demand of the transformer area as the constraint, a reactive power allocation optimization model is established. An uncertainty parameter and a two-stage decision-making framework are introduced: the reactive power demand of the transformer area is modeled as a bounded uncertain parameter that varies within a certain interval, and a two-stage robust optimization decision-making framework is adopted. In the first stage, the switching state of the discrete equipment capacitors is decided, and in the second stage, after the uncertainty scenario is revealed, the reactive power output of the static var compensator is continuously adjusted. Model Solving and Strategy Generation: The two-stage robust optimization model is solved iteratively using a column constraint generation algorithm. The main problem outputs the capacitor switching strategy, and the sub-problem identifies the worst reactive power demand scenario. Accelerated computation by leveraging an industrial-grade configuration platform: Through the multi-configuration parallel computing architecture of the industrial-grade configuration platform, different reactive power demand scenarios are modularly divided and computed in parallel, reducing the iteration time of model solving and improving computational efficiency.

[0006] Preferably, the process of constructing the reactive power allocation optimization model includes: Within the transformer area, local voltage regulation is achieved through capacitors and static var compensators. The reactive power compensation amount is allocated according to the regulation cost of the two types of reactive power compensation devices to determine the switching status of capacitors and the reactive power output of static var compensators. The total cost of the objective function includes the sum of capacitor switching loss cost and static var compensator (SVC) adjustment cost; the constraints include upper and lower limits of SVC reactive power output, power balance constraint that the total reactive power compensation is not less than the reactive power demand of the transformer area, and integer constraint that the capacitor switching state is 0 or 1.

[0007] Preferably, the process of introducing uncertainty parameters and a two-stage decision-making framework includes: The range of reactive power demand for the transformer area is determined based on the reactive power demand forecast or benchmark value, and the range is defined by the maximum possible forecast deviation. The reason for choosing the capacitor switching state as the decision object in the first stage is that capacitors use mechanical switches, have slow operation speed and are not suitable for frequent operation, so it is necessary to make a decision in advance that does not depend on uncertain scenarios. The reason for choosing the reactive power output of the static var compensator as the decision object in the second stage is that the static var compensator has the characteristic of continuous adjustment and can respond and adjust in real time after the uncertainty scenario is revealed.

[0008] Preferably, the iterative solution process using the column constraint generation algorithm includes: Initialization: Set the number of iterations to 0, the lower bound of the objective function to negative infinity, the upper bound to positive infinity, and the initial scene set to an empty set; Solve the main problem: Optimize the capacitor switching state based on the known scenario, output the current optimal capacitor switching strategy and the objective function value of the main problem, and update the lower bound of the objective function; Solve the subproblem: With the capacitor switching strategy output by the main problem fixed, find the worst reactive power demand scenario for the transformer area and the minimum adjustment cost of the static var compensator under this scenario, output the objective function value of the subproblem and the worst reactive power demand scenario, and update the upper bound of the objective function. Constraint addition and convergence judgment: The worst reactive power demand scenario obtained from the subproblem is added to the main problem in the form of constraints. The number of iterations is increased and the main problem and subproblems are solved repeatedly until the difference between the upper and lower bounds of the objective function is less than the preset threshold. The iteration stops and the final capacitor switching strategy and the real-time output command of the static var compensator are output.

[0009] Preferably, the specific methods for accelerating computation based on an industrial-grade configuration platform include: Parallel scenario allocation: The reactive power demand scenarios of different transformer areas are allocated to independent computing nodes of the industrial-grade configuration platform, and the sub-problems corresponding to each scenario are solved simultaneously. Asynchronous communication and collaboration: An asynchronous communication mechanism is used between the main problem and sub-problems to avoid the time loss of the main problem waiting for the sub-problems to be solved, and to reduce the overall iteration waiting time. Parallel Integer Programming: For the 0-1 integer programming problem of capacitor switching states, a parallelized branch and bound method is used to accelerate the solution.

[0010] A reactive power distribution acceleration system for power distribution networks based on an industrial-grade configuration platform is disclosed, including: The model building module is used to establish a reactive power allocation optimization model with the objective function of minimizing the total cost of reactive power regulation in the transformer area and the constraint of meeting the reactive power demand of the transformer area. The uncertainty handling and decision-making module is used to model the reactive power demand of the transformer area as a bounded uncertain parameter within a certain interval, and to construct a two-stage robust optimization decision framework. The first stage decides the switching state of the capacitor, and the second stage decides the real-time output of the static var compensator. The column constraint generation and solution module is used to iteratively solve the two-stage robust optimization model using a column constraint generation algorithm. It outputs the capacitor switching strategy by solving the main problem and identifies the worst reactive power demand scenario by solving the sub-problems. The industrial-grade configuration parallel acceleration module is used to modularize and perform parallel computing on different reactive power demand scenarios based on the multi-configuration parallel computing architecture of the industrial-grade configuration platform, thereby accelerating the computation of model solving.

[0011] Preferably, the model building module includes: Parameter input unit: Used to input parameters of the reactive power compensation device and basic data on reactive power demand in the transformer area; Objective function generation unit: used to generate an objective function that minimizes the total cost based on the capacitor switching loss cost and the static var compensator adjustment cost; Constraint generation unit: used to generate upper and lower limit constraints for static var compensator output, reactive power balance constraints, and integer constraints for capacitor switching states.

[0012] Preferably, the industrial-grade configuration parallel acceleration module includes: Scenario segmentation unit: used to modularly split different uncertain scenarios of reactive power demand in transformer areas; Node allocation unit: Used to allocate the split scene to independent computing nodes of the industrial-grade configuration platform; Asynchronous communication unit: used to establish asynchronous communication links between the main problem-solving module and the sub-problem-solving module, reducing iteration waiting time; Integer Programming Parallel Unit: Used to solve 0-1 integer programming problems involving capacitor switching states using a parallelized branch and bound method.

[0013] Preferably, it also includes a visualization module, which is used to display the calculation results. The displayed content includes the reactive power output allocation results of each reactive power compensation device, a reactive power compensation cost composition analysis chart, a dynamic change chart of capacitor switching status, a system reactive power supply and demand balance chart, a voltage change chart of each node, and an economic index chart.

[0014] Preferably, it also includes a real-time data interaction module, which is used to collect real-time reactive power demand data of the transformer area and transmit it to the column constraint generation and solution module. At the same time, it sends the real-time output command of the static var compensator obtained by the solution to the static var compensator device, and collects the actual switching status of the capacitor and feeds it back to the model building module to correct and optimize the model.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The core improvement of this invention lies in accelerating the computation of reactive power allocation models for power distribution networks by relying on the multi-configuration parallel computing architecture of an industrial-grade configuration platform. With the help of the industrial-grade configuration platform, real-time performance and robustness can be significantly improved. Simultaneously, the optimization results can be dynamically displayed through visualization components, generating multi-dimensional analytical views such as equipment output, cost structure, and voltage stability, supporting engineering implementation. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the reactive power allocation method for transformer substations according to an embodiment of the present invention. Figure 2 The diagram shows the switching state and reactive power of the static compensator obtained in the embodiment. Detailed Implementation

[0017] Please refer to Figures 1 to 2 The present invention relates to a method and system for accelerating reactive power allocation calculation in power distribution networks based on an industrial-grade configuration platform, comprising: Constructing a reactive power allocation optimization model: With minimizing the total cost of reactive power regulation in the transformer area as the objective function and meeting the reactive power demand of the transformer area as the constraint, a reactive power allocation optimization model is established. Uncertainty parameters and decision-making framework are introduced: the reactive power demand of the transformer area is modeled as a bounded uncertain parameter that varies within a certain interval, and a two-stage robust optimization decision-making framework is adopted; in the first stage, the switching state of discrete equipment capacitors is decided, and in the second stage, after the uncertainty scenario is revealed, the reactive power output of the static var compensator is continuously adjusted. Model Solving and Strategy Generation: The two-stage robust optimization model is solved iteratively using a column constraint generation algorithm. The main problem outputs the capacitor switching strategy, and the sub-problem identifies the worst reactive power demand scenario. Through iteration, a robust optimization scheme that is feasible under all possible uncertain scenarios is finally obtained, realizing the economic allocation of reactive power in the transformer area.

[0018] Furthermore, the process of constructing the reactive power allocation optimization model is as follows: Within the transformer area, local voltage regulation can be achieved using capacitors or static var compensators (SVCs). The reactive power allocation strategy is to distribute the reactive power compensation amount based on the regulation cost of the reactive power compensation device, thereby obtaining the switching status of the capacitors and the reactive power output of the SVCs. The objective function and constraints are shown in the following equation: ; ; Where: i is the reactive power compensation device number; C is the set of static compensator numbers; U is the set of capacitor numbers; pi is the adjustment cost of reactive power compensation device i; Qi is the reactive power of static compensator i; Qc,i is the reactive power of capacitor i, used as a constant input model; xi is the switching state of capacitor i; QD is the amount of reactive power that the transformer area needs to generate.

[0019] The constraint that controls the reactive power output of the static compensator to not exceed the threshold is shown in the following formula: ; Where: Qi,min and Qi,max are the lower and upper limits of the reactive power output of static compensator i; Since reactive power compensation is required, the total compensation power must be greater than the required reactive power QD, as shown in the following formula: ; Furthermore, uncertainty parameters and decision-making frameworks are introduced, including: To further consider the impact of QD on control, in actual transformer substations, especially with a high proportion of photovoltaic and electric vehicle integration, QD is a highly uncertain parameter. Deterministic optimization results may be completely ineffective in the face of actual fluctuations, or even lead to voltage overruns or frequent equipment malfunctions. Therefore, a robust optimization method is introduced, which no longer assumes precise knowledge of QD, but acknowledges its uncertainty and assumes that it varies within a defined "uncertainty set": ; in: This refers to the predicted or baseline value of reactive power demand. For the maximum possible prediction deviation Meanwhile, since the objective function cannot perfectly match the constantly changing QD, a two-stage robust optimization framework is used to model this problem. The decision-making process is divided into two stages. The first stage involves decisions that must be made in advance and are independent of uncertainty. Because the mechanical switching of capacitors is slow and not suitable for frequent operation, the switching state of the capacitors is the first-stage decision. The second-stage decision is one that can be quickly adjusted after uncertainty is revealed. Since the reactive power output of the static compensator is continuously adjustable and can respond in real time to the actual value of the reactive power demand QD, it is used as the second-stage decision.

[0020] The objective function is modified as follows: ; Where: the outer layer min is the optimal switching combination of the capacitor bank. To minimize the total cost; the middle layer max is for each fixed... Uncertainty QD in its set Choose the worst-case scenario for the inner layer to maximize the total cost (i.e., the worst system operating condition); the inner layer uses a given combination of capacitors. Under the worst-case reactive power demand QD, optimize the output of the static compensator. To minimize the operating costs in this scenario.

[0021] After reorganizing the above formulas, the final optimized model for charging pile power adjustment is as follows: ; Furthermore, the process of model solving and policy generation includes: The model was solved using a column constraint generation algorithm, which involves iteratively solving the objective function by dividing it into a main problem and sub-problems. The process consists of the following steps: Step 1: Solving the main problem and deciding on capacitor switching And a temporary optimal solution is given.

[0022] Step Two: Subproblems: Given the main problem The goal is to find the worst-case QD and its corresponding scheduling scheme for the static compensator. The subproblem is a bi-level optimization problem, which is transformed into a single-level maximization problem using duality theory before being solved.

[0023] Step 3: Add the worst-case scenario QD and its corresponding constraints found in the subproblems to the main problem, and solve the main problem again.

[0024] Step 4: Repeat steps 1, 2, and 3 until convergence to the robust optimal solution. The solution to the objective function is then obtained.

[0025] Step 5: Based on the real-time reactive power demand data, and on the basis of the switched capacitor banks, optimize the calculation of the real-time reactive power output command of the static var compensator.

[0026] Furthermore, the industrial-grade configuration platform accelerates computation in the following ways: (1) Assign different reactive power demand scenarios (QD) to independent computing nodes to solve subproblems in parallel; (2) The main problem and subproblems adopt an asynchronous communication mechanism to reduce iteration waiting time; (3) The 0-1 integer programming of the capacitor switching strategy is implemented in parallel using the branch and bound method.

[0027] Furthermore, the computation results are fully displayed through the visualization components of the industrial-grade configuration platform: (1) The reactive power distribution results of each compensation device; (2) Analysis chart of reactive power compensation cost composition; (3) Dynamic change diagram of capacitor switching state; (4) Reactive power supply and demand balance diagram of the system; (5) Voltage variation diagram at each node; (6) Economic indicators chart.

[0028] To verify the effectiveness and superiority of the reactive power allocation optimization model proposed in this invention, we conducted simulation analysis based on a distribution substation example from a real-world case. The design of the example followed the principles of typicality, reproducibility, and rationality, and a solution and visualization program was developed based on an industrial-grade configuration platform.

[0029] This example includes the following elements: Compensation equipment: It is equipped with 3 sets of parallel capacitors and 2 static var compensators, forming a typical "discrete + continuous" hybrid reactive power compensation system.

[0030] Equipment parameters: 1. Capacitor bank: its rated capacity The compensation levels were set to 100kVar, 150kVar, and 200kVar, respectively, covering different compensation gradients. The unit loss cost was set to 0.8, 1.0, and 1.2 yuan / kVar, reflecting the economic characteristic that the larger the capacity, the higher the unit switching loss cost.

[0031] 2. Static Var Compensator (SVC): Its reactive power regulation range is set at [-50, 100] kVar and [-80, 120] kVar, reflecting the SVC's bidirectional regulation capability of both generating and absorbing reactive power. Its unit regulation cost (1.5-2.0 yuan / kVar) is higher than that of a capacitor, which is consistent with the objective fact that its operation and maintenance costs are higher as a precision, continuous regulation device.

[0032] 3. Load demand: Average reactive power demand of the transformer area. Set to 250kVar, fluctuation range The requirement is 50kVar. This level of demand matches the total capacity of the configured equipment, which not only reflects the necessity of optimized allocation but also effectively verifies the model's solution capability.

[0033] The code generates comprehensive visualization charts, including: reactive power allocation results, which intuitively show the reactive power output undertaken by each compensation device (capacitor and SVC); cost composition, which analyzes the proportion of capacitor loss and SVC regulation cost in the total cost; capacitor switching status diagram, which clearly shows the switching status of each capacitor bank; reactive power supply and demand balance comparison, which compares the total reactive power supply and demand to verify whether the optimization results meet the constraints; voltage stability analysis, which simulates the effect of reactive power compensation on the improvement of node voltage and intuitively shows the improvement of voltage quality by optimization; and economic indicator analysis, which shows key economic indicators such as total cost and unit reactive power cost.

[0034] By solving the above examples, the following optimization results were obtained, all of which demonstrate good rationality and economy: (1) Switching state of capacitors The result is [0,1,1] or similar 0-1 integer combinations, which perfectly matches the integer constraints in the model, proving that the solver effectively handles discrete decision problems; (2) Reactive power output of SVC The result (e.g., [80.5, 69.5] kVar) strictly falls within its upper and lower bound constraints. , Within [the specified range], the validity of the constraints was verified; (3) The total cost consists of capacitor switching losses and SVC adjustment costs. The results show that the adjustment cost of SVC is usually higher than that of capacitor. This is entirely in line with expectations, because the goal of the model is to minimize the total cost. The optimization algorithm will prioritize the use of capacitors with lower unit cost for coarse adjustment, and then use SVCs with higher unit cost but more flexible adjustment for fine adjustment. This cost structure clearly reflects the economic optimization principle pursued by this invention.

[0035] (4) The reactive power balance constraint is strictly satisfied, that is, the total reactive power supply ≥ reactive power demand. In the calculation results, the total supply is usually slightly greater than the demand (e.g., 258kVar > 250kVar), which not only ensures the technical requirements of voltage stability, but also reflects the characteristic of the optimization algorithm to pursue economy under the safety constraints.

[0036] (5) Voltage stability analysis shows that after reactive power compensation, the voltage at each node is near the rated value (10kV) and has even increased. Voltage change and reactive power output The positive correlation is consistent with the basic principle of power system "reactive power compensation supports voltage", proving the technical correctness of the optimization results.

[0037] (6) The results of this model provide a good solution and benchmark for solving more complex robust optimization models, proving the practicality and scalability of the method of this invention.

[0038] The above implementation process fully demonstrates the technical details, parameter settings, data processing flow and system integration method of the present invention in multimodal remote sensing semantic change detection, and verifies its effectiveness in mitigating catastrophic forgetting and improving generalization ability and interpretability.

[0039] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for accelerating reactive power distribution in power distribution networks based on an industrial-grade configuration platform, characterized in that, Includes the following steps: Constructing a reactive power allocation optimization model: With minimizing the total cost of reactive power regulation in the transformer area as the objective function and meeting the reactive power demand of the transformer area as the constraint, a reactive power allocation optimization model is established. An uncertainty parameter and a two-stage decision-making framework are introduced: the reactive power demand of the transformer area is modeled as a bounded uncertain parameter that varies within a certain interval, and a two-stage robust optimization decision-making framework is adopted. In the first stage, the switching state of the discrete equipment capacitors is decided, and in the second stage, after the uncertainty scenario is revealed, the reactive power output of the static var compensator is continuously adjusted. Model Solving and Strategy Generation: The two-stage robust optimization model is solved iteratively using a column constraint generation algorithm. The main problem outputs the capacitor switching strategy, and the sub-problem identifies the worst reactive power demand scenario. Accelerated computation by leveraging an industrial-grade configuration platform: Through the multi-configuration parallel computing architecture of the industrial-grade configuration platform, different reactive power demand scenarios are modularly divided and computed in parallel, reducing the iteration time of model solving and improving computational efficiency.

2. The method for accelerating reactive power distribution in a distribution network based on an industrial-grade configuration platform according to claim 1, characterized in that, The process of constructing the reactive power allocation optimization model includes: Within the transformer area, local voltage regulation is achieved through capacitors and static var compensators. The reactive power compensation amount is allocated according to the regulation cost of the two types of reactive power compensation devices to determine the switching status of capacitors and the reactive power output of static var compensators. The total cost of the objective function includes the sum of capacitor switching loss cost and static var compensator (SVC) adjustment cost; the constraints include upper and lower limits of SVC reactive power output, power balance constraint that the total reactive power compensation is not less than the reactive power demand of the transformer area, and integer constraint that the capacitor switching state is 0 or 1.

3. The method for accelerating reactive power distribution in a distribution network based on an industrial-grade configuration platform according to claim 1 or 2, characterized in that, The process of introducing uncertainty parameters and a two-stage decision-making framework includes: The range of reactive power demand for the transformer area is determined based on the predicted or benchmark value of reactive power demand, and the range is defined by the maximum possible prediction deviation. The reason for choosing the capacitor switching state as the decision object in the first stage is that capacitors use mechanical switches, have slow operation speed and are not suitable for frequent operation, so it is necessary to make a decision in advance that does not depend on uncertain scenarios. The reason for choosing the reactive power output of the static var compensator as the decision object in the second stage is that the static var compensator has the characteristic of continuous adjustment and can respond and adjust in real time after the uncertainty scenario is revealed.

4. The method for accelerating reactive power distribution in a distribution network based on an industrial-grade configuration platform according to claim 1, characterized in that, The iterative solution process using the column constraint generation algorithm includes: Initialization: Set the number of iterations to 0, the lower bound of the objective function to negative infinity, the upper bound to positive infinity, and the initial scene set to an empty set; Solve the main problem: Optimize the capacitor switching state based on the known scenario, output the current optimal capacitor switching strategy and the objective function value of the main problem, and update the lower bound of the objective function; Solve the subproblem: With the capacitor switching strategy output by the main problem fixed, find the worst reactive power demand scenario for the transformer area and the minimum adjustment cost of the static var compensator under this scenario, output the objective function value of the subproblem and the worst reactive power demand scenario, and update the upper bound of the objective function. Constraint addition and convergence judgment: The worst reactive power demand scenario obtained from the subproblem is added to the main problem in the form of constraints. The number of iterations is increased and the main problem and subproblems are solved repeatedly until the difference between the upper and lower bounds of the objective function is less than the preset threshold. The iteration stops and the final capacitor switching strategy and the real-time output command of the static var compensator are output.

5. The method for accelerating reactive power distribution in a distribution network based on an industrial-grade configuration platform according to claim 1, characterized in that, The specific methods for accelerating computation based on an industrial-grade configuration platform include: Parallel scenario allocation: The reactive power demand scenarios of different transformer areas are allocated to independent computing nodes of the industrial-grade configuration platform, and the sub-problems corresponding to each scenario are solved simultaneously. Asynchronous communication and collaboration: An asynchronous communication mechanism is used between the main problem and sub-problems to avoid the time loss of the main problem waiting for the sub-problems to be solved, and to reduce the overall iteration waiting time. Parallel Integer Programming: For the 0-1 integer programming problem of capacitor switching states, a parallelized branch and bound method is used to accelerate the solution.

6. A system for implementing the method for accelerating reactive power distribution in a power distribution network based on an industrial-grade configuration platform as described in any one of claims 1-5, characterized in that, include: The model building module is used to establish a reactive power allocation optimization model with the objective function of minimizing the total cost of reactive power regulation in the transformer area and the constraint of meeting the reactive power demand of the transformer area. The uncertainty handling and decision-making module is used to model the reactive power demand of the transformer area as a bounded uncertain parameter within a certain interval, and to construct a two-stage robust optimization decision framework. The first stage decides the capacitor switching state, and the second stage decides the real-time output of the static var compensator. The column constraint generation and solution module is used to iteratively solve the two-stage robust optimization model using a column constraint generation algorithm. It outputs the capacitor switching strategy by solving the main problem and identifies the worst reactive power demand scenario by solving the sub-problems. The industrial-grade configuration parallel acceleration module is used to modularize and perform parallel computing on different reactive power demand scenarios based on the multi-configuration parallel computing architecture of the industrial-grade configuration platform, thereby accelerating the computation of model solving.

7. The power distribution network reactive power distribution acceleration system based on an industrial-grade configuration platform according to claim 6, characterized in that, The model building module includes: Parameter input unit: Used to input parameters of the reactive power compensation device and basic data on reactive power demand in the transformer area; Objective function generation unit: used to generate an objective function that minimizes the total cost based on the capacitor switching loss cost and the static var compensator adjustment cost; Constraint generation unit: used to generate upper and lower limit constraints for static var compensator output, reactive power balance constraints, and integer constraints for capacitor switching states.

8. The power distribution network reactive power distribution acceleration system based on an industrial-grade configuration platform according to claim 6, characterized in that, The industrial-grade configuration parallel acceleration module includes: Scenario segmentation unit: used to modularly split different uncertain scenarios of reactive power demand in transformer areas; Node allocation unit: Used to allocate the split scene to independent computing nodes of the industrial-grade configuration platform; Asynchronous communication unit: used to establish asynchronous communication links between the main problem-solving module and the sub-problem-solving module, reducing iteration waiting time; Parallel Integer Programming Unit: Used to solve 0-1 integer programming problems involving capacitor switching states using a parallelized branch and bound method.

9. The power distribution network reactive power distribution acceleration system based on an industrial-grade configuration platform according to claim 6, characterized in that, It also includes a visualization module, which is used to display the calculation results. The displayed content includes the reactive power output allocation results of each reactive power compensation device, reactive power compensation cost composition analysis chart, dynamic change chart of capacitor switching status, system reactive power supply and demand balance chart, voltage change chart of each node, and economic index chart.

10. The power distribution network reactive power distribution acceleration system based on an industrial-grade configuration platform according to claim 6, characterized in that, It also includes a real-time data interaction module, which is used to collect real-time reactive power demand data of the transformer area and transmit it to the column constraint generation and solution module. At the same time, it sends the real-time output command of the static var compensator obtained by the solution to the static var compensator device, and collects the actual switching status of the capacitor and feeds it back to the model building module to correct and optimize the model.

Images