Power distribution network inversion calculation method and system considering dynamic partition and topological state, and medium

Abstract

The embodiment of the invention discloses a power distribution network inversion calculation method and system considering a dynamic partition and a topological state and a medium, and the method comprises the steps: obtaining an actual measurement value of the electrical quantity of a power distribution network in a typical operation state through a preset measurement action, and constructing an initial covariance matrix of a preset to-be-estimated parameter according to the actual measurement value; and inputting the to-be-estimated parameter as a sample into a preset power flow model to obtain an electric quantity analog measurement value. Performing regional division on the power distribution network to calculate a posterior covariance matrix of each region, and combining the initial covariance matrix to obtain information gain of each region. An area with the information gain exceeding a preset threshold value is determined as a key area, a sparse constraint optimization function is established based on the switch-on state of the area and a to-be-estimated parameter, and a parameter estimation value and a switch-on estimation value are obtained through inversion; and if the information gain does not exceed the threshold value, adopting a historical inversion result of a switch conduction state and a to-be-estimated parameter. And through dynamic partitioning and information gain screening, accurate inversion of the topology and parameters of the power distribution network is realized.

Description

Technical Field

[0001] This invention relates to the field of power distribution network technology, and in particular to a power distribution network inversion calculation method, system and medium that considers dynamic partitioning and topology status. Background Technology

[0002] With the deepening of smart grid construction, transparent operation and state awareness of distribution networks have become key means to improve power supply reliability and optimize operational efficiency. Currently, research on the transparency of distribution networks mainly relies on local electrical quantity data provided by various measurement devices, and uses methods such as state estimation and power flow inversion to infer the operating state of the power grid.

[0003] However, existing methods still have the following significant shortcomings when dealing with complex distribution network structures, variable operating conditions, and equipment aging: First, existing inversion methods mostly use independent inversion of electrical quantities, failing to effectively combine topology state information, resulting in the inversion process being disconnected from the actual physical state of the equipment. This makes it difficult to capture small deviations in line parameters and potential operational risks, leading to low reliability of early warnings and high false alarm and false alarm rates. Secondly, in terms of inversion region processing strategies, most methods still adopt a fixed partitioning approach, which lacks flexibility and adaptability. They cannot dynamically adjust the computational granularity according to the value of real-time information, resulting in insufficient estimation accuracy in key regions and wasted computational resources in non-key regions. Furthermore, existing methods generally fail to correlate electrical condition estimation with actual equipment health status, economic losses, and other factors, lacking the ability to make decisions with both physical and economic transparency. Although some studies have attempted to introduce auxiliary monitoring methods, such as point cloud comparison and vibration alarms, these often only capture macroscopic changes and fail to deeply integrate with the electrical structure and power flow characteristics of the power grid, resulting in limited discriminative capabilities and an inability to support economically driven operational decisions. Summary of the Invention

[0004] Based on this, it is necessary to propose a distribution network inversion calculation method, system, and medium that considers dynamic partitioning and topology status to address the above problems.

[0005] A distribution network inversion calculation method considering dynamic partitioning and topology status, the method comprising: The actual measured values ​​of electrical quantities in the distribution network under typical operating conditions are obtained according to the preset measurement actions, and the initial covariance matrix of the preset parameters to be estimated is constructed based on the actual measured values.

[0006] The preset parameters to be estimated are used as parameter samples and input into the preset power flow model to output the corresponding simulated measurement values ​​of the electrical quantities of the distribution network.

[0007] The power distribution network is divided into regions, and the posterior covariance matrix of each region is obtained.

[0008] The information gain for each region is calculated based on the initial covariance matrix and the posterior covariance matrix.

[0009] Regions with information gain greater than a preset threshold are designated as key regions. A sparse constraint optimization function is established based on the switch conduction state and preset parameters to be estimated within the key regions. The parameter estimates and switch conduction estimates are then derived from the sparse constraint optimization function and used as the topology state and parameters of the key regions.

[0010] The region with information gain less than or equal to a preset threshold is designated as a non-critical region. The historical inversion results of the switch conduction state and preset estimated parameters of the non-critical region are used as the topology state and parameters of the non-critical region.

[0011] Specifically, the step of using preset parameters to be estimated as parameter samples, inputting them into a preset power flow model, and outputting the corresponding simulated measured values ​​of the distribution network electrical quantities includes: A power flow model is established based on the laws of circuit physics to characterize the quantitative relationship between electrical quantities in the distribution network and preset parameters to be estimated.

[0012] The preset parameters to be estimated are used as parameter samples.

[0013] The power distribution network is divided into several regions, and the parameter samples of each region are substituted into the power flow model to determine the corresponding simulated measurement values.

[0014] Specifically, calculating the information gain of each region based on the initial covariance matrix and the posterior covariance matrix includes: according to Calculate the information gain for each region, where, The posterior covariance matrix is... Let the initial covariance matrix be... For information gain.

[0015] Specifically, the step of establishing a sparse constraint optimization function based on the switch conduction state within the region and a preset parameter to be estimated includes: according to Establish a sparse-constrained optimization function, where, For simulated measurement values, This is the sum of squared errors between the simulated and actual measured values. The switch is in the ON state. For sparsity constraints, The relaxation hardening coefficient is... These are actual measured values. These are simulated measurement values.

[0016] The method further includes: The residual vector and residual covariance are calculated based on the parameter estimates and the switch conduction estimates. The reliability of the inversion is then evaluated based on the residual vector and residual covariance.

[0017] The process of calculating the residual vector and residual covariance based on the parameter estimates and switch conduction estimates, constructing the Fisher information matrix, and evaluating the inversion reliability based on the Fisher information matrix further includes: The residual vector is sparsely represented using a dictionary learning method to obtain a sparse coefficient vector.

[0018] The parameter estimates are compensated based on the sparse vectors to obtain corrected parameter estimates.

[0019] The step of compensating the parameter estimate based on the sparse vector to obtain the corrected parameter estimate further includes: Obtain candidate measurement actions.

[0020] A utility value is constructed based on the ratio of the square of the information gain to the measurement cost of the candidate measurement action.

[0021] If the utility value is greater than the gain threshold, a new candidate measurement action is added to obtain the actual measured value of the electrical quantity of the distribution network under typical operating conditions.

[0022] If the utility value is less than or equal to the gain threshold, the final parameter estimate and switch conduction estimate are output based on the comparison between the L2 norm of the residual vector and the tolerance threshold.

[0023] Specifically, the step of outputting the final parameter estimate and switch conduction estimate based on the comparison between the L2 norm of the residual vector and the tolerance threshold includes: Determine the L2 norm of the residual vector.

[0024] If the L2 norm of the residual vector is greater than or equal to the preset threshold, then return to the step of establishing a sparse constraint optimization function based on the switch conduction state in the region and the preset parameters to be estimated.

[0025] If the L2 norm of the residual vector is less than the preset threshold, then the final parameter estimate and switch conduction estimate are output.

[0026] A distribution network inversion calculation system considering dynamic partitioning and topology status, the system comprising: The data acquisition module is used to acquire the actual measured values ​​of electrical quantities of the distribution network under typical operating conditions according to preset measurement actions, and to construct an initial covariance matrix of preset parameters to be estimated based on the actual measured values.

[0027] The information gain determination module is used to input the preset parameters to be estimated as parameter samples into the preset power flow model and output the corresponding simulated measurement values ​​of the electrical quantities of the distribution network; divide the distribution network into regions and obtain the posterior covariance matrix of each region; and calculate the information gain of each region based on the initial covariance matrix and the posterior covariance matrix.

[0028] The topology state and parameter inversion module is used to designate regions with information gain greater than a preset threshold as critical regions, establish a sparse constraint optimization function based on the switch conduction state and preset parameters to be estimated within the critical regions, and invert the parameter estimates and switch conduction estimates through the sparse constraint optimization function to serve as the topology state and parameters of the critical regions; and designate regions with information gain less than or equal to the preset threshold as non-critical regions, and use the historical inversion results of the switch conduction state and preset parameters to be estimated in the non-critical regions as the topology state and parameters of the non-critical regions.

[0029] A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method described above.

[0030] The embodiments of the present invention have the following beneficial effects: This invention combines switch conduction states with preset parameters to be estimated, making the inversion calculation results closer to the actual operating state of the power grid. This overcomes the problem of disconnect between electrical quantity inversion and topological state in existing technologies, achieving synchronous estimation of distribution network structure and equipment parameters, and significantly improving the physical consistency and accuracy of state perception. Furthermore, this invention uses information gain to dynamically divide key regions, automatically triggering refined modeling and inversion in these regions, significantly improving local estimation accuracy while avoiding wasting computational resources in low-information-value areas, reducing unnecessary computational resource consumption, and obtaining more accurate estimation results in key regions, thus improving the overall system performance and computational efficiency. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] in: Figure 1 This is a flowchart illustrating an embodiment of a distribution network inversion calculation method considering dynamic partitioning and topology status provided by the present invention. Figure 2A flowchart illustrating another embodiment of the distribution network inversion calculation method considering dynamic partitioning and topology status provided by the present invention; Figure 3 A schematic diagram of an embodiment of a distribution network inversion calculation system considering dynamic partitioning and topology state provided by the present invention; Figure 4 A schematic diagram of the structure of an embodiment of the device provided by the present invention; Figure 5 A schematic diagram of the structure of an embodiment of the medium provided by the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] like Figure 1 As shown, Figure 1 This is a flowchart illustrating an embodiment of a distribution network inversion calculation method considering dynamic partitioning and topology state provided by the present invention. The method includes: S101: Obtain the actual measured values ​​of electrical quantities in the distribution network under typical operating conditions according to the preset measurement actions, and construct the initial covariance matrix of the preset parameters to be estimated based on the actual measured values.

[0035] For example, actual measurements of electrical quantities in the distribution network under typical operating conditions are collected according to preset measurement actions, including voltage, current, power flow, and switch operation records. For each preset parameter to be estimated, such as line resistance, reactance, or transformer ratio, its prior variance is constructed, and an initial covariance matrix is ​​estimated by combining it with historical data to quantify the uncertainty of the parameter.

[0036] Specifically, the initial covariance matrix can be expressed by the following formula: ; In the formula, Indicates parameters The prior variance; This represents the first parameter in the distribution network parameters to be estimated. These parameters include, but are not limited to, line parameters (such as line resistance). and reactance), transformer parameters (such as transformer turns ratio) The selection of parameters such as load (e.g., load power factor) depends on the object to be inverted. For example, if the goal is to estimate the line conditions of the entire distribution network, then... This includes the resistance and reactance of each line within the network; if the goal is to estimate the topology or load of critical nodes, then... It also includes switch status or load parameters.

[0037] Corresponding prior variance Indicates parameters The magnitude of the uncertainty is used to quantify the potential range of fluctuation of the parameter without introducing new measurement actions. For example, if... The resistance of a certain line can be statistically analyzed by examining the current, voltage, and power relationships of that line during its historical operation, combined with the accuracy of the equipment, to obtain the variance of the resistance fluctuation. ;if If it is the transformer turns ratio, its variance can be set based on the nameplate accuracy and historical load variations.

[0038] It should be noted that by collecting actual measurement data under typical operating conditions and constructing the prior distribution and initial covariance matrix of the parameters to be estimated, the uncertainty of the parameters can be scientifically quantified, providing reliable initial conditions for subsequent inversion calculations.

[0039] S102: Input the preset parameters to be estimated as parameter samples into the preset power flow model, and output the corresponding simulated measurement values ​​of the electrical quantities of the distribution network.

[0040] For example, based on circuit laws (Kirchhoff's laws) and component characteristics, a quantitative relationship is established between electrical measurements and preset parameters to be estimated, resulting in a preset power flow model. The preset parameters to be estimated are used as parameter samples. The inputs to the preset power flow model are parameter samples x (line resistance, reactance, etc.) and switch conduction states z. The output is the analog measured values ​​of electrical quantities in the distribution network. The power flow model can be represented by S(x,z).

[0041] S103: Divide the distribution network into regions and obtain the posterior covariance matrix for each region.

[0042] For example, the distribution network is divided into multi-scale regions, primarily based on measurement information density supplemented by electrical distance and topology. Specifically, for each region, the posterior covariance matrix is ​​determined according to the formula shown below. : ; In the formula, For the sample size, Let x be the mean of the parameter samples.

[0043] S104: Calculate the information gain for each region based on the initial covariance matrix and the posterior covariance matrix.

[0044] For example, the information gain of each region is calculated based on the initial covariance matrix and the posterior covariance matrix, and it is determined whether the mesh needs to be further refined to improve the local inversion accuracy. The information gain is: ; In the formula, The posterior covariance matrix is... Let the initial covariance matrix be... For information gain.

[0045] It should be noted that by dividing the distribution network into multi-scale regions and using the information gain index to evaluate the contribution of each region to the inversion accuracy, the need for grid refinement is dynamically determined. This step enables automatic triggering of refined calculations in key areas, improving local estimation accuracy, while avoiding wasting computational resources in low-contribution areas.

[0046] Furthermore, a preset threshold is set. (e.g., 0.1).

[0047] S105: The region with information gain greater than the preset threshold is taken as the key region. A sparse constraint optimization function is established based on the switch conduction state and preset parameters to be estimated in the key region. The parameter estimate and switch conduction estimate are obtained by inverting the sparse constraint optimization function and used as the topology state and parameters of the key region.

[0048] For example, when When this happens, the system automatically triggers refinement and parameter inversion in that region. In this way, the system can adaptively improve parameter estimation accuracy in key regions while avoiding wasting computational resources in low-contribution areas. Specifically, it assumes the parameter prior covariance matrix... Assuming that the posterior covariance is obtained after adding the measurement... ,but , Calculate information gain: = The regions with information gain greater than a preset threshold are designated as key regions and then finely divided.

[0049] Furthermore, a sparse constraint optimization function is established based on the switch conduction state and preset parameters to be estimated within the key region. The parameter estimates and switch conduction estimates are then derived from the sparse constraint optimization function and used as the topology state and parameters of the key region.

[0050] Specifically, a least-squares optimization framework based on power flow equations is adopted. A sparse constraint optimization function is established based on the switch conduction state in the key region and the preset parameters to be estimated. The sparse constraint optimization function is shown in the following equation: ; in, For simulated measurement values, This is the sum of squared errors between simulated and actual measured values, used to approximate the actual system state. When the switch is on, it is initially a continuous variable of 0–1. To impose sparsity constraints on the topology, switching states are encouraged to change only when necessary. The relaxation hardening coefficient can be gradually increased with iterations to balance minimizing error and topological sparsity. These are actual measured values, such as voltage, current, and power data provided by the PMU system. These are simulated measurement values.

[0051] Further, initialization Initialize to the prior mean or the result of the previous inversion. Set the regularization coefficient for historical switch states or all-one vectors. Initial value.

[0052] Furthermore, perform alternating iterative optimization: Fixed topology Optimize parameters : ; Solve using the gradient descent algorithm; Fixed topology Optimize parameters : ; Optimization is performed using a proximal gradient algorithm with sparse constraints.

[0053] Iterate until convergence, and output the parameter estimates. and the estimated value of the switch conduction state .

[0054] It should be noted that regional refinement mainly refers to further subdividing the original power grid regional model, which is divided into initial regions, into smaller node and branch units, and re-establishing the power flow equations and parameter variable sets at this subdivision scale. For example, a long line can be re-divided into sub-units of 100 meters, and resistance, reactance, and load distribution parameters can be established separately; for complex nodes, the refinement is further refined to independent variables of switches, branches, and branch loads. Subsequently, joint topology (switch conduction state) – parameter inversion is performed on these key regions, making the local parameter estimation more accurate and able to capture local anomalies and subtle inconsistencies that are difficult to detect under the initial regional division.

[0055] S106: Regions with information gain less than or equal to a preset threshold are designated as non-critical regions. The historical inversion results of the switch-on state and preset parameters to be estimated in the non-critical regions are used as the topology state and parameters of the non-critical regions.

[0056] For example, when When the region is considered a non-critical region, the historical inversion results of the switch conduction state and preset parameters to be estimated in the non-critical region are used as the topology state and parameters of the non-critical region.

[0057] It should be noted that the switch conduction state in this invention refers to the switch conduction rate (SCR). The aim is to transform the discrete topology state identification problem into a continuously solvable optimization problem, while simultaneously achieving continuous monitoring of the health status of switching equipment. Its physical meaning is: a closed switch corresponds to... (Extremely low on-resistance), disconnect the corresponding When a switch exhibits abnormalities such as contact aging, contact surface corrosion, or mechanical jamming, its contact resistance increases and conductivity decreases. In this case, the estimated switch conduction state obtained through inversion will fall within the open interval (0,1), and its value continuously characterizes the degree of deterioration in the health state; specifically, a lower value indicates poorer conduction performance and a more severe abnormality. After inversion convergence, a preset threshold (usually 0.5) can be used to... The topology is discretized into 0 / 1 states, but its continuous estimates are retained as an important basis for equipment condition assessment. This method achieves an inherent unity between topology identification and equipment health monitoring, enhancing the depth and physical interpretability of condition perception. Furthermore, the preset threshold can be determined based on a large amount of historical data. The system supports fine-tuning of this threshold according to the actual equipment type (such as load switches and circuit breakers) to adapt to different scenarios. Moreover, there is no need to set a threshold individually for each switch; a unified threshold can meet most application requirements, avoiding the problem of overly complex parameters.

[0058] As described above, this invention, by combining switch conduction states with preset parameters to be estimated, makes the inversion calculation results closer to the actual operating state of the power grid. This overcomes the problem of disconnect between electrical quantity inversion and topological state in existing technologies, achieving synchronous estimation of distribution network structure and equipment parameters, and significantly improving the physical consistency and accuracy of state perception. Furthermore, this invention uses information gain to dynamically divide key regions, automatically triggering refined modeling and inversion in these regions, significantly improving local estimation accuracy. Simultaneously, it avoids wasting computational resources in low-information-value areas, reducing unnecessary computational resource consumption. More accurate estimation results can be obtained in key regions, improving the overall system performance and computational efficiency.

[0059] like Figure 2 As shown, Figure 2 This is a flowchart illustrating another embodiment of the distribution network inversion calculation method considering dynamic partitioning and topology state provided by the present invention. The distribution network inversion calculation method considering dynamic partitioning and topology state includes: S201: Obtain the actual measured values ​​of electrical quantities in the distribution network under typical operating conditions according to the preset measurement actions, and construct the initial covariance matrix of the preset parameters to be estimated based on the actual measured values.

[0060] S202: The preset parameters to be estimated are used as parameter samples and input into the preset power flow model to output the corresponding simulated measurement values ​​of the electrical quantities of the distribution network.

[0061] S203: Divide the distribution network into regions and obtain the posterior covariance matrix for each region.

[0062] S204: Calculate the information gain for each region based on the initial covariance matrix and the posterior covariance matrix.

[0063] S205: The region with information gain greater than the preset threshold is taken as the key region. A sparse constraint optimization function is established based on the switch conduction state and preset parameters to be estimated in the key region. The parameter estimate and switch conduction estimate are inverted through the sparse constraint optimization function and used as the topology state and parameters of the key region.

[0064] S206: Regions with information gain less than or equal to a preset threshold are designated as non-critical regions. The switch-on state of the non-critical regions and the historical inversion results of the preset parameters to be estimated are used as the topology state and parameters of the non-critical regions.

[0065] It should be noted that steps S201-S206 are in Figure 1 The implementation scenarios shown have been discussed in detail and will not be repeated here.

[0066] S207: Calculate the residual vector and residual covariance based on the parameter estimates and switch conduction estimates, and evaluate the reliability of the inversion based on the residual vector and residual covariance.

[0067] For example, based on the parameter estimates and the switch conduction estimates, the difference between the actual measured data and the model output is used as the residual vector: ; In the formula, These are parameter estimates. This is the estimated value for switch conduction.

[0068] Subsequently, the residual covariance is calculated to reflect the uncertainty of the estimate: ; In the formula, Let be the covariance matrix of the residuals, representing the statistical correlation and volatility of the residuals across different measurement dimensions. This represents the total number of samples, typically between 100 and 1000. Indicates the first A residual vector, This represents the mean vector of all sample residuals, reflecting the average deviation of the residuals. Indicates the first The outer product of the deviations of each residual from the mean is used to characterize the correlation between residuals in each dimension.

[0069] The overall significance of the formula for calculating residual covariance lies in calculating the covariance matrix of multiple independent residual samples through statistical analysis. This is to characterize the overall uncertainty of the model in the measurement dimension. If A large diagonal element indicates significant fluctuations in the residuals of some measurements, reflecting unreliable parameter or topological estimations. Conversely, if the off-diagonal elements are significantly non-zero, it suggests strong correlations between different measurements, indicating potential structural biases in the model. Therefore, It is not only used to quantitatively describe the distribution characteristics of inversion error, but also provides a basis for subsequent uncertainty quantification and proactive measurement decisions.

[0070] Simultaneously, a Fisher information matrix is ​​constructed to quantify uncertainty:

[0071] In the formula, For the power flow equations with respect to parameters The Jacobian matrix can be obtained by automatic differentiation; The noise covariance matrix can be calibrated by the sensor accuracy.

[0072] It should be noted that this step assesses the statistical characteristics and uncertainty of the estimation error by calculating the residual vector and its covariance matrix between the actual measurements and the model output. Furthermore, a Fisher information matrix is ​​constructed to quantify the reliability of the parameter estimation, providing a basis for subsequent correction and decision-making.

[0073] S208: Use dictionary learning to sparsely represent the residual vector to obtain a sparse coefficient vector.

[0074] S209: Compensate the parameter estimates based on the sparse vectors to obtain the corrected parameter estimates.

[0075] For example, a dictionary learning method is used to sparsely represent the residuals: ; In the formula, It is the residual vector; The residual dictionary matrix can be obtained through offline training using the OMP algorithm; It is a sparse coefficient vector.

[0076] The sparse coefficient vector can be represented by the following formula: ; In the formula, is the regularization coefficient.

[0077] Furthermore, the parameter estimates are compensated based on the sparse vectors to obtain the corrected parameter estimates, as shown in the following equation: ; In the formula, For the corrected parameter estimates; The mapping matrix is ​​obtained through offline training using historical residual data.

[0078] Specifically, the residual dictionary matrix is ​​set. sparse coefficient vector Mapping matrix Preliminary parameter estimation ,but: ; .

[0079] As can be seen from the above, through sparse representation and mapping correction, parameter estimation can be improved from... Optimized to The corrected parameters are closer to the true values, demonstrating the effectiveness of residual correction.

[0080] It should be noted that this step employs a dictionary learning method to sparsely represent the residuals, and corrects the parameter estimates through sparse coefficient vectors, thereby improving the accuracy and robustness of the inversion results. This step achieves effective residual compensation through an offline-trained residual dictionary and mapping matrix.

[0081] S210: Obtain candidate measurement actions.

[0082] S211: Construct a utility value based on the ratio of the square of the information gain to the measurement cost of the candidate measurement action.

[0083] S2111: If the utility value is greater than the gain threshold, a new candidate measurement action is added to obtain the actual measured value of the electrical quantity of the distribution network under typical operating conditions.

[0084] S2112: If the utility value is less than or equal to the gain threshold, the final parameter estimate and switch conduction estimate are output based on the comparison between the L2 norm of the residual vector and the tolerance threshold.

[0085] For example, candidate measurement actions are obtained. A utility value is then constructed based on the ratio of the square of the information gain to the measurement cost of the candidate measurement action. ; In the formula, For candidate measurement actions The utility value, The information gain resulting from the execution of the action. The cost of the action.

[0086] Among them, action cost The evaluation breaks down the implementation costs of candidate measurement actions into multiple sub-items, including equipment and material costs, labor costs, time costs, communication and data processing costs, energy costs, risk losses, and opportunity costs. These sub-items are then weighted and aggregated according to preset weights to obtain a unified cost value. The weights can be determined through expert experience. The data for each sub-item can come from equipment purchase contracts, work order records, operational statistics, or risk probability models. When necessary, different dimensions are converted into monetary equivalents through conversion methods. This forms a quantitative indicator that can objectively reflect the cost of measurement deployment and can be compared with the cost-effectiveness of information gain, providing a basis for the selection and optimization of proactive measurement actions. Action Cost The costs are uniformly quantified into monetary costs through a standardized conversion framework, with specific calculations automatically based on preset system parameters: For example, equipment and material costs are directly taken from the unit price and quantity of equipment in the asset database; labor costs are determined by multiplying the standard labor rate database by the work hour quota; time costs are obtained by multiplying the unit time value (based on the system's average power outage loss or operating cost) by the action delay time; communication and data processing costs are calculated based on data transmission volume and publicly available cloud computing service tariffs; energy costs are derived by multiplying equipment power consumption, operating time, and electricity price; risk losses are the product of historically statistical risk occurrence probabilities and corresponding economic losses; opportunity costs are estimated based on the amount of capital tied up and the system's average return on investment. All sub-items are automatically aggregated in monetary units to form an objective and unified scalar value. It supports the automation and standardization of closed-loop decision-making without the need for manual empowerment.

[0087] Furthermore, set a gain threshold. (e.g., 1), if Then, the newly added candidate measurement action can obtain the actual measured values ​​of electrical quantities of the distribution network under typical operating conditions.

[0088] It should be noted that this step constructs an active measurement utility function, which integrates information gain and implementation cost to decide whether to perform a new measurement. If the utility value exceeds a threshold, a new candidate measurement action is added to obtain the actual measured values ​​of electrical quantities in the distribution network under typical operating conditions; otherwise, it is determined whether the residual value has converged, forming a closed-loop optimization mechanism.

[0089] like Set tolerance threshold (Usually set to 2–3 times the measurement accuracy), when the L2 norm of the residual vector When this happens, a rollback to step S205 is triggered for re-analysis; When the residual vector satisfies the 2-norm At that time, the final parameter estimate and switch conduction estimate are output to achieve closed-loop iteration.

[0090] It should be noted that this step sets a residual tolerance threshold to determine whether the inversion results converge. If the condition is met, the final parameters and topology state are output and the historical database is updated; otherwise, a rollback and reanalysis are triggered to achieve closed-loop iteration and continuous system optimization.

[0091] As described above, this invention deeply integrates electrical condition estimation with equipment health status and economic loss factors to construct a "physical-economic" dual transparency system. It also introduces a utility function based on the ratio of information gain to cost to achieve economically driven adaptive measurement deployment and closed-loop decision-making, thereby enhancing the pertinence and effectiveness of power grid operation management.

[0092] Furthermore, by combining sparse constraint optimization with dictionary learning residual compensation mechanism, the ability to identify small deviations in line parameters and potential faults is effectively enhanced, significantly improving the reliability of fault warning, while greatly reducing the risk of false alarms and missed alarms, and improving the robustness of the system under complex operating conditions.

[0093] Furthermore, relying on the measurement of uncertainty and the closed-loop rollback mechanism, the system can adaptively trigger supplementary measurements or rollback optimization under conditions of measurement anomalies or data loss, possessing strong fault tolerance and adaptive capabilities, and supporting the high-reliability operation and continuous optimization of the distribution network in a variable environment.

[0094] like Figure 3 As shown, Figure 3 This is a schematic diagram of an embodiment of a distribution network inversion calculation system considering dynamic partitioning and topology state provided by the present invention. A distribution network inversion calculation system 10 considering dynamic partitioning and topology state, the system includes: The data acquisition module 11 is used to acquire the actual measured values ​​of electrical quantities of the distribution network under typical operating conditions according to the preset measurement actions, and to construct the initial covariance matrix of the preset parameters to be estimated based on the actual measured values.

[0095] The information gain determination module 12 is used to input the preset parameters to be estimated as parameter samples into the preset power flow model and output the corresponding analog measurement values ​​of the electrical quantities of the distribution network; divide the distribution network into regions and obtain the posterior covariance matrix of each region; and calculate the information gain of each region based on the initial covariance matrix and the posterior covariance matrix.

[0096] The topology state and parameter inversion module 13 is used to identify regions with information gain greater than a preset threshold as key regions, establish a sparse constraint optimization function based on the switch conduction state and preset parameters to be estimated within the key regions, and invert parameter estimates and switch conduction estimates through the sparse constraint optimization function to serve as the topology state and parameters of the key regions; and identify regions with information gain less than or equal to the preset threshold as non-key regions, and use the historical inversion results of the switch conduction state and preset parameters to be estimated in the non-key regions as the topology state and parameters of the non-key regions.

[0097] For example, in the data acquisition module 11, actual measured values ​​of the electrical quantities of the distribution network under typical operating conditions are acquired according to preset measurement actions, and an initial covariance matrix of the preset parameters to be estimated is constructed based on the actual measured values. In the information gain determination module 12, a power flow model characterizing the quantitative relationship between the electrical quantities of the distribution network and the preset parameters to be estimated is first established based on the laws of circuit physics; the preset parameters to be estimated are used as parameter samples; the distribution network is divided into several regions, and the parameter samples of each region are substituted into the power flow model to determine the corresponding simulated measured values. Further, the distribution network is divided into regions, and the posterior covariance matrix of each region is obtained; the information gain of each region is calculated based on the initial covariance matrix and the posterior covariance matrix. In the topology state and parameter inversion module 13, regions with information gain greater than a preset threshold are designated as critical regions. A sparse constraint optimization function is established based on the switch conduction state and preset parameters to be estimated within the critical regions. The parameter estimates and switch conduction estimates are then inverted through the sparse constraint optimization function and used as the topology state and parameters of the critical regions. Regions with information gain less than or equal to the preset threshold are designated as non-critical regions. The historical inversion results of the switch conduction state and preset parameters to be estimated in the non-critical regions are used as the topology state and parameters of the non-critical regions.

[0098] Distribution network inversion calculation systems that consider dynamic zoning and topology states also include: The evaluation module is used to calculate the residual vector and residual covariance based on the parameter estimates and the switch conduction estimates, and to evaluate the reliability of the inversion based on the residual vector and residual covariance.

[0099] The parameter estimate correction module is used to perform sparse representation of the residual vector using a dictionary learning method to obtain a sparse coefficient vector. The parameter estimates are then compensated based on the sparse coefficient vector to obtain corrected parameter estimates.

[0100] A new measurement action module is added to acquire candidate measurement actions. A utility value is constructed based on the ratio of the square of the information gain to the measurement cost of the candidate measurement action. If the utility value is greater than the gain threshold, a new candidate measurement action is added to acquire the actual measured values ​​of the electrical quantities of the distribution network under typical operating conditions. If the utility value is less than or equal to the gain threshold, the final parameter estimate and switch conduction estimate are output based on the comparison between the L2 norm of the residual vector and the tolerance threshold.

[0101] like Figure 4 As shown, Figure 4 This is a schematic diagram of an embodiment of the device provided by the present invention. The device 20 includes a memory 21 and a processor 22. The memory 21 stores a computer program, and the processor 22 executes the computer program during operation to achieve, for example... Figure 1 and Figure 2 The method shown.

[0102] The specific technical details of the distribution network inversion calculation method that considers dynamic partitioning and topology state when the above-mentioned device 20 executes the computer program have been discussed in detail in the above method steps, so they will not be repeated here.

[0103] like Figure 5 As shown, Figure 5 This is a schematic diagram of the structure of an embodiment of the medium provided by the present invention. The medium 30 stores at least one computer program 31, which is executed by the processor 22 to perform the following... Figure 1 and Figure 2 The method shown is detailed above and will not be repeated here. In one embodiment, the medium 30 can be a storage chip, hard disk, portable hard disk, USB flash drive, optical disk, or other read / write storage device, or even a server, etc.

[0104] Furthermore, the processes depicted in the accompanying drawings do not necessarily have to be performed in the specific or sequential order shown to achieve the desired result. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0105] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer-readable storage media are basically similar to the method embodiments, and therefore described more simply; relevant parts can be referred to the descriptions of the method embodiments.

[0106] The apparatus, device, non-volatile computer-readable storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, device and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, device and non-volatile computer storage medium will not be repeated here.

[0107] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0108] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components. Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of computer program products 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.

[0109] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. 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, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0110] 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.

[0111] 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.

[0112] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0113] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0114] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0115] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0116] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0117] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0118] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A distribution network inversion calculation method considering dynamic partitioning and topology status, characterized in that, The method includes: The actual measured values ​​of electrical quantities in the distribution network under typical operating conditions are obtained according to the preset measurement actions, and the initial covariance matrix of the preset parameters to be estimated is constructed based on the actual measured values. The preset parameters to be estimated are used as parameter samples and input into the preset power flow model to output the corresponding simulated measurement values ​​of the electrical quantities of the distribution network. The power distribution network is divided into regions, and the posterior covariance matrix of each region is obtained. Calculate the information gain for each region based on the initial covariance matrix and the posterior covariance matrix; The region with information gain greater than a preset threshold is designated as the key region. A sparse constraint optimization function is established based on the switch conduction state and preset parameters to be estimated within the key region. The parameter estimate and switch conduction estimate are then derived from the sparse constraint optimization function and used as the topology state and parameters of the key region. The region with information gain less than or equal to a preset threshold is designated as a non-critical region. The historical inversion results of the switch conduction state and preset estimated parameters of the non-critical region are used as the topology state and parameters of the non-critical region.

2. The distribution network inversion calculation method considering dynamic zoning and topology status according to claim 1, characterized in that, The step of using preset parameters to be estimated as parameter samples, inputting them into a preset power flow model, and outputting the corresponding simulated measurement values ​​of distribution network electrical quantities specifically includes: A power flow model is established based on the laws of circuit physics to characterize the quantitative relationship between electrical quantities in the distribution network and preset parameters to be estimated. The preset parameters to be estimated are used as parameter samples; The power distribution network is divided into several regions, and the parameter samples of each region are substituted into the power flow model to determine the corresponding simulated measurement values.

3. The distribution network inversion calculation method considering dynamic zoning and topology status according to claim 1, characterized in that, The calculation of the information gain for each region based on the initial covariance matrix and the posterior covariance matrix specifically includes: according to Calculate the information gain for each region, where, For the posterior covariance matrix, The initial covariance matrix, For information gain.

4. The distribution network inversion calculation method considering dynamic partitioning and topology status according to claim 1, characterized in that, The step of establishing a sparse constraint optimization function based on the switch conduction state within the region and preset parameters to be estimated specifically includes: according to Establish a sparse-constrained optimization function, where, For simulated measurement values, This is the sum of squared errors between the simulated and actual measured values. The switch is in the ON state. For sparsity constraints, The relaxation hardening coefficient is... These are actual measured values. These are simulated measurement values.

5. The distribution network inversion calculation method considering dynamic zoning and topology status according to claim 1, characterized in that, The method further includes: The residual vector and residual covariance are calculated based on the parameter estimates and the switch conduction estimates. The reliability of the inversion is then evaluated based on the residual vector and residual covariance.

6. The distribution network inversion calculation method considering dynamic zoning and topology status according to claim 5, characterized in that, The process of calculating the residual vector and residual covariance based on the parameter estimates and switch conduction estimates, constructing the Fisher information matrix, and evaluating the inversion reliability based on the Fisher information matrix further includes: The residual vector is sparsely represented using a dictionary learning method to obtain a sparse coefficient vector. The parameter estimates are compensated based on the sparse vectors to obtain corrected parameter estimates.

7. The distribution network inversion calculation method considering dynamic zoning and topology status according to claim 6, characterized in that, After compensating the parameter estimate based on the sparse vector to obtain the corrected parameter estimate, the method further includes: Obtain candidate measurement actions; A utility value is constructed based on the ratio of the square of the information gain to the measurement cost of the candidate measurement action; If the utility value is greater than the gain threshold, a new candidate measurement action is added to obtain the actual measured value of the electrical quantity of the distribution network under typical operating conditions; If the utility value is less than or equal to the gain threshold, the final parameter estimate and switch conduction estimate are output based on the comparison between the L2 norm of the residual vector and the tolerance threshold.

8. The distribution network inversion calculation method considering dynamic partitioning and topology status according to claim 7, characterized in that, The step of outputting the final parameter estimate and switch conduction estimate based on the comparison between the L2 norm of the residual vector and the tolerance threshold specifically includes: Determine the L2 norm of the residual vector; If the L2 norm of the residual vector is greater than or equal to the preset threshold, then return to the step of establishing a sparse constraint optimization function based on the switch conduction state in the region and the preset parameters to be estimated. If the L2 norm of the residual vector is less than the preset threshold, then the final parameter estimate and switch conduction estimate are output.

9. A distribution network inversion calculation system considering dynamic partitioning and topology status, characterized in that, The system includes: The data acquisition module is used to acquire the actual measured values ​​of electrical quantities of the distribution network under typical operating conditions according to preset measurement actions, and to construct an initial covariance matrix of preset parameters to be estimated based on the actual measured values. The information gain determination module is used to input the preset parameters to be estimated as parameter samples into the preset power flow model and output the corresponding simulated measurement values ​​of the electrical quantities of the distribution network; divide the distribution network into regions and obtain the posterior covariance matrix of each region; and calculate the information gain of each region based on the initial covariance matrix and the posterior covariance matrix. The topology state and parameter inversion module is used to designate regions with information gain greater than a preset threshold as critical regions, establish a sparse constraint optimization function based on the switch conduction state and preset parameters to be estimated within the critical regions, and invert the parameter estimates and switch conduction estimates through the sparse constraint optimization function to serve as the topology state and parameters of the critical regions; and designate regions with information gain less than or equal to the preset threshold as non-critical regions, and use the historical inversion results of the switch conduction state and preset parameters to be estimated in the non-critical regions as the topology state and parameters of the non-critical regions.

10. A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method as claimed in any one of claims 1 to 8.

Images