A power distribution network state estimation method, system, device, medium and product

Abstract

The application discloses a power distribution network state estimation method, system, device, medium and product, relates to the power grid state monitoring technical field, obtains the power grid topology structure of the power distribution network, carries out partition processing on the power grid topology structure, obtains a plurality of sub-regions, obtains the measurement data of each sub-region, respectively carries out state evaluation on each measurement data based on a preset least square objective function, obtains a plurality of initial state data, respectively carries out basic state prediction according to each measurement data and each initial state data, obtains a plurality of partition state estimations, carries out state correction on each partition state estimation, and obtains the corresponding power distribution network state estimation. The technical problem that the existing power distribution network state estimation is solved. The recursive least square algorithm is used, but the forgetting factor of the recursive least square algorithm is fixed, cannot adapt to the sudden change of new energy output, is prone to estimation lag, large error and other problems, and the accuracy of the power distribution network state estimation is reduced.

Description

Technical Field

[0001] This invention relates to the field of power grid condition monitoring technology, and in particular to a method, system, device, medium and product for estimating the condition of a distribution network. Background Technology

[0002] With the large-scale integration of distributed power sources such as photovoltaics, wind power, energy storage, and flexible controllable loads into the distribution network, traditional passive distribution networks are gradually transforming into distributed smart distribution networks with multi-source coupling and complex operating characteristics. The grid operation exhibits new characteristics such as random power fluctuations, bidirectional power flow, and widening peak-valley load differences, placing higher demands on real-time situational awareness, safety management, and dispatch optimization of the distribution network. Dynamic state estimation, as the core foundational technology for distribution network situational awareness, relies on real-time electrical data collected by multi-source measurement devices such as SCADA, micro PMUs, and smart meters. It uses numerical algorithms to solve for core state variables such as node voltage and branch power across the entire network. It serves as a prerequisite support for distribution network fault early warning, power flow control, reactive power optimization, and safety verification. The accuracy, real-time performance, and robustness of the estimation results directly determine the reliability of distribution network operation and management.

[0003] Currently, most existing distribution network state estimation methods use recursive least squares algorithms. However, recursive least squares algorithms have a fixed forgetting factor, which cannot adapt to sudden changes in the output of new energy sources. They are prone to problems such as estimation lag and large errors, which reduce the accuracy of distribution network state estimation. Summary of the Invention

[0004] This invention provides a method, system, device, medium, and product for estimating the state of a distribution network. It solves the technical problem that existing distribution network state estimation methods mostly use recursive least squares algorithms, but the forgetting factor of recursive least squares algorithms is fixed, which cannot adapt to the sudden change in the output of new energy sources. They are prone to problems such as estimation lag and large errors, which reduce the accuracy of distribution network state estimation.

[0005] The first aspect of this invention provides a method for estimating the state of a distribution network, comprising:

[0006] Obtain the power grid topology of the distribution network, and partition the power grid topology to obtain multiple sub-regions;

[0007] Measurement data of each sub-region is acquired, and state evaluation of each measurement data is performed based on a preset least squares objective function to obtain multiple initial state data.

[0008] Based on each of the measurement data and the initial state data, basic state prediction is performed to obtain multiple partition state estimates.

[0009] The state estimates of each partition are corrected to obtain the corresponding distribution network state estimates.

[0010] Optionally, the step of performing state evaluation on each of the measurement data based on a preset least squares objective function to obtain multiple initial state data includes:

[0011] Each measurement data point is used to construct a corresponding measurement vector.

[0012] The preset least squares objective function is solved iteratively using each of the aforementioned measurement vectors to obtain multiple initial state data.

[0013] Optionally, the step of performing basic state prediction based on each of the measurement data and each of the initial state data to obtain multiple partition state estimates includes:

[0014] The corresponding output measurement sequence and input measurement sequence are constructed using each of the aforementioned measurement data;

[0015] The output measurement sequence and input measurement sequence corresponding to each of the measurement data are respectively input into a preset first least squares estimation function to obtain multiple parameter vectors;

[0016] The preset second least squares estimation function is iteratively solved using the parameter vectors and initial state data corresponding to each of the measurement data to obtain multiple partition state estimates.

[0017] Optionally, the step of performing state correction on each of the partition state estimates to obtain the corresponding distribution network state estimate includes:

[0018] Principal component analysis was used to decompose the state estimates of each partition, resulting in multiple signal data and multiple noise data.

[0019] Random noise simulation operations are performed on each of the aforementioned noise data to obtain multiple simulated noise data.

[0020] Each of the aforementioned signal data and its corresponding analog noise data is summed to obtain multiple analog measurement data.

[0021] Orthogonal least squares fitting was performed on each of the simulated measurement data to obtain multiple fitting parameters;

[0022] Based on a preset number of random simulations, the fitting parameters corresponding to the state estimates of each partition are cyclically sampled and statistically analyzed to obtain the corresponding power distribution network state estimates.

[0023] Optionally, the step of cyclically sampling and statistically analyzing the fitting parameters corresponding to the state estimates of each partition based on a preset number of random simulations to obtain the corresponding distribution network state estimates includes:

[0024] When the number of fitting parameters is greater than or equal to the preset number of random simulations, the fitting parameters corresponding to each partition state estimate are averaged to obtain multiple fitting average values.

[0025] Calculate the standard deviation of the fitting parameters corresponding to each partition state estimate to obtain multiple fitting standard deviations;

[0026] The average value and standard deviation of each of the aforementioned fitting values ​​are used as the corresponding power distribution network state estimates;

[0027] When the number of fitting attempts for the fitting parameters is less than the number of random simulations, the process jumps to the step of performing random noise simulation operations based on each of the noise data to obtain multiple simulated noise data.

[0028] Optionally, the second least squares estimation function is specifically:

[0029] ;

[0030] ;

[0031] ;

[0032] ;

[0033] in, For the partition state estimation at time t, These are the estimated values ​​of the basic parameters at time t. For time indexing, Let be the covariance matrix at time t. This is the deviation compensation coefficient. These are the estimated values ​​of the basic parameters at time t-1. Let be the corrected gain vector at time t. The output measurement value at time t. Let be the information vector at time t. Let be the covariance matrix at time t-1. Forgetting factor, It is an identity matrix.

[0034] A second aspect of the present invention provides a power distribution network state estimation system, comprising:

[0035] The acquisition module is used to obtain the power grid topology of the distribution network, and to partition the power grid topology to obtain multiple sub-regions;

[0036] The evaluation module is used to acquire measurement data of each of the sub-regions, and to evaluate the state of each measurement data based on a preset least squares objective function to obtain multiple initial state data.

[0037] The prediction module is used to perform basic state prediction based on each of the measurement data and each of the initial state data to obtain multiple partition state estimates.

[0038] The correction module is used to correct the state estimates of each partition to obtain the corresponding distribution network state estimates.

[0039] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the power distribution network state estimation method as described above.

[0040] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the power distribution network state estimation method as described above.

[0041] The fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer performs the power distribution network state estimation method as described above.

[0042] As can be seen from the above technical solutions, the present invention has the following advantages:

[0043] This invention acquires the topology of the distribution network, partitions the topology to obtain multiple sub-regions, obtains measurement data for each sub-region, and performs state evaluation on each measurement data based on a preset least squares objective function to obtain multiple initial state data. Based on each measurement data and each initial state data, a basic state prediction is performed to obtain multiple sub-regional state estimates. Finally, state corrections are applied to each sub-regional state estimate to obtain the corresponding distribution network state estimate. This invention overcomes the technical problem that existing distribution network state estimation methods often use recursive least squares algorithms, which have a fixed forgetting factor, cannot adapt to sudden changes in renewable energy output, and are prone to estimation lag and large errors, thus reducing the accuracy of distribution network state estimation. Compared with traditional distribution network state estimation methods, this invention first decomposes a large-scale distribution network into multiple small sub-regions, significantly reducing the matrix operation dimension of a single state solution and effectively reducing the computing power consumption of a single node, thus meeting the timeliness requirements of real-time estimation for distributed distribution networks. Second, based on local measurement data of each region, the initial state is independently solved through a least-squares objective function, making full use of the redundancy of local measurements in each region and avoiding the measurement timing mismatch problem caused by global data transmission delay. The initial state results can quickly complete local coarse estimation, providing a reliable benchmark for time series prediction. Basic state prediction is performed based on each measurement data and each initial state data, breaking through the shortcomings of traditional static state estimation that relies only on single-moment measurements and cannot track the fluctuations of new energy sources and loads. It relies on the continuous characteristics of time series to capture the dynamic changes of the power grid caused by sudden changes in photovoltaic and wind power output, and outputs a regional state estimate that conforms to the time series evolution law, solving the problems of state lag and poor change tracking ability of traditional algorithms. Finally, state correction is performed on all regional state estimates to obtain an accurate distribution network state estimate, thereby improving the accuracy of distribution network state estimation. Attached Figure Description

[0044] 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.

[0045] Figure 1 This is a flowchart illustrating the steps of a power distribution network state estimation method provided in Embodiment 1 of the present invention.

[0046] Figure 2 A flowchart illustrating the steps of a power distribution network state estimation method provided in Embodiment 2 of the present invention.

[0047] Figure 3 This is a comparison diagram of the state estimation effects of the method of this application and the conventional method provided in Embodiment 2 of the present invention;

[0048] Figure 4 This is a structural block diagram of a power distribution network state estimation system provided in Embodiment 3 of the present invention;

[0049] Figure 5 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0050] This invention provides a method, system, device, medium, and product for estimating the state of a distribution network. It addresses the technical problem that existing distribution network state estimation methods often employ recursive least squares algorithms, which have a fixed forgetting factor and cannot adapt to sudden changes in new energy output. These algorithms are prone to problems such as estimation lag and large errors, thus reducing the accuracy of distribution network state estimation.

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0052] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a power distribution network state estimation method provided in Embodiment 1 of the present invention.

[0053] This invention provides a method for estimating the state of a distribution network, comprising:

[0054] Step 101: Obtain the power grid topology of the distribution network, and partition the power grid topology to obtain multiple sub-regions.

[0055] The power grid topology refers to the physical architecture model consisting of all primary equipment such as buses (nodes), lines (branches), transformers, and switches in the distribution network and their electrical connections. It is usually stored in the form of node-branch diagrams or undirected graphs to characterize the actual flow paths of current and power in the power grid, as well as the electrical distance and coupling strength between components.

[0056] A subregion refers to a local subnetwork unit of the distribution network obtained after partitioning. Each subregion contains several internal nodes and branches, as well as at least one boundary node shared with adjacent regions. The subregion has sufficient measurement redundancy to support independent execution of state estimation calculations, while also achieving coordinated operation with the entire distribution network through state information interaction with the boundary nodes.

[0057] In this embodiment of the invention, the power grid topology of the distribution network is obtained, and the power grid topology is partitioned. Based on the feeder routing, geographical distribution, voltage level, and location of distributed power source access points, the entire distribution network is divided into multiple mutually coupled but relatively independent sub-regions. When partitioning, it is necessary to ensure that each sub-region has sufficient measurement redundancy to support subsequent independent state estimation calculations. At the same time, common boundary nodes are retained as inter-regional coordination variables so that the consistency and coordination of global estimation results can be achieved through boundary information interaction, resulting in multiple sub-regions that can independently perform state estimation calculations.

[0058] It is worth mentioning that by acquiring the complete distribution network topology and dividing it into multiple independent sub-regions, the computational task of the entire distribution network, which is large in scale and has a large number of nodes and branches, can be decoupled and split. This significantly reduces the matrix dimension of a single state estimation solution, effectively reduces the computational load of a single computing device, and significantly improves the overall computing speed by relying on parallel computing in multiple sub-regions, meeting the real-time requirements of dynamic state estimation in distribution networks with high renewable energy penetration. The partitioning is based on feeders, electrical coupling relationships, and measurement observability. Each sub-region has close electrical connections and low cross-regional coupling. When conducting independent local calculations, there is weak mutual interference between regions. At the same time, each sub-region only needs to exchange a small amount of boundary node data, without uploading all local topology and measurement raw data, thus balancing the data privacy and information exchange efficiency of each region. In addition, after partitioning, each sub-region can independently complete the entire process of measurement acquisition, state assessment, and time series prediction. When there are measurement gaps or equipment disturbances in the local power grid, only the calculation results of the corresponding sub-region are affected, without causing interruption of the overall state estimation calculation, thus improving the stability of the algorithm.

[0059] Step 102: Obtain measurement data for each sub-region, and perform state evaluation on each measurement data based on the preset least squares objective function to obtain multiple initial state data.

[0060] Measurement data refers to the electrical quantitative information that is collected in real time by sensors such as voltage transformers, current transformers and power measurement devices installed at each node of the distribution network and output after analog-to-digital conversion. Specifically, it covers node voltage amplitude, node voltage phase angle, branch active power, branch reactive power, node injected active power and node injected reactive power, etc.

[0061] Initial state data refers to the initial estimates of the voltage magnitude and phase angle of nodes within each sub-region obtained by iteratively solving the least squares objective function.

[0062] In this embodiment of the invention, the measurement data uploaded in real time by the local acquisition terminal of each sub-region is retrieved, and each measurement data is input into a preset least squares objective function to construct the objective function of each sub-region. The node voltage and branch power flow calculation results corresponding to the minimization condition of the objective function of each sub-region are solved. During the calculation, the range of state quantity values ​​is limited according to the complete topological constraints within the sub-region to avoid abnormal solutions such as voltage exceeding limits and power imbalance caused by unconstrained solutions. After all sub-regions are solved in parallel, the node voltage and branch power flow values ​​output by all sub-regions are summarized to obtain multiple initial state data.

[0063] It should be noted that the least squares objective function is as follows:

[0064] ;

[0065] ;

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] in, The objective function value, For measurement vectors, For state vectors, The measurement function describes the nonlinear mapping relationship between the state vector and the measurement vector. The measurement noise covariance matrix is ​​a diagonal matrix. The state quantity at time k is predicted by the state transition function from the state estimate at time k-1. This is the state transition function. Let be the state error covariance prediction matrix at time k. The state transition matrix (i.e., the state transition function in) Jacobian matrix at the location). Let k be the state error covariance prediction matrix at time k-1. The process noise covariance matrix describes the inherent uncertainties of the system's dynamic process, such as random load fluctuations and changes in distributed power output. Let K be the filter gain matrix (Kalman gain) at time k. Let be the measurement Jacobian matrix at time k, which is the measurement function at the predicted state value. The linearized matrix, Let k be the measurement noise covariance matrix at time k. Let be the posterior state estimate at time k, which is the final state estimate after the measurement update at the current time (incorporating the new measurement information). The predicted state value (prior state estimate) at time k is the state quantity at time k predicted using the final state estimate at time k-1 through the state transition function. Let k be the measurement vector at time k. It is an identity matrix with the same dimensions as the state vector, where diagonal elements are 1s and off-diagonal elements are 0s. Let be the state error covariance matrix obtained after updating at time k, where k is the time index.

[0071] The state transition function is as follows:

[0072] ;

[0073] ;

[0074] ;

[0075] in, This is the predicted state value at time k+1. Let k be the horizontal component. Let be the tilt component at time k. The first smoothing parameter is... The state prediction value at time k. The second smoothing parameter, The tilt component at time k-1, This represents the horizontal component at time k-1.

[0076] It is worth mentioning that the least squares objective function takes minimizing the weighted sum of squares of measurement residuals as its core optimization. This can fully utilize redundant measurement data within the partition to suppress the calculation bias caused by random noise from sensors, and quickly solve for the initial solutions of node voltages and branch power flows that fit the actual operating conditions of the region, providing a reliable benchmark for subsequent time-series state prediction. Each partition conducts state assessment independently and is isolated from each other. Measurement anomalies or data loss in a single region will only affect the initial state results of that region and will not cause global calculation failure, thus improving the overall robustness of the algorithm. At the same time, independent partition solutions do not need to consider cross-regional electrical coupling constraints, simplifying the single optimization solution model, reducing matrix operation complexity, and quickly outputting multiple sets of decoupled initial state data. This provides a layered and lightweight original calculation foundation for subsequent basic state prediction and noise simulation correction, balancing the accuracy of initial value estimation, computational efficiency, and system operation stability.

[0077] Step 103: Perform basic state prediction based on each measurement data and each initial state data to obtain multiple partition state estimates.

[0078] Partition state estimation refers to the refined dynamic estimation results obtained after each sub-region has undergone the basic state prediction process. This result integrates the temporal evolution information of the measurement data, the static reference information of the initial state data, and the dynamic characteristic information of the system obtained by dynamic parameter identification. It includes the optimal estimated values ​​of the node voltage amplitude and phase angle at the current moment, and also implies the prediction information of the state change trend at the next moment.

[0079] In this embodiment of the invention, corresponding output measurement sequences and input measurement sequences are constructed using various measurement data. The output and input measurement sequences corresponding to each measurement data are then input into a preset first least squares estimation function to solve for multiple parameter vectors that optimally describe the dynamic relationship between the input and output of a sub-region. Using the parameter vectors corresponding to each measurement data and initial state data as initial conditions, a preset second least squares estimation function is iteratively solved time-by-time to obtain multiple partition state estimates.

[0080] It is worth mentioning that by performing basic state prediction based on each measurement data and each initial state data, multiple partition state estimates are obtained. This fully leverages the significant advantages of dynamic time series calculation compared to traditional static estimation. This method is no longer limited to instantaneous measurement information at a single moment, but combines historical measurement time series change patterns with the initial state obtained in the previous step to build a dynamic deduction logic. It can accurately capture the continuous changes in grid state caused by random fluctuations in distributed photovoltaic and wind power output and load peak-valley switching, effectively improving the shortcomings of traditional static state estimation, which cannot track the dynamic trend of the grid and has a serious estimation lag when the state changes abruptly.

[0081] Step 104: Perform state correction on the state estimates of each zone to obtain the corresponding distribution network state estimates.

[0082] Distribution network status estimation refers to the final estimate of the overall network operating status obtained after three core processing steps: independent calculation by region, dynamic recursive prediction, and Monte Carlo statistical correction. The result is based on the voltage amplitude and phase angle of each node, and is accompanied by the corresponding estimation uncertainty index. It can provide grid dispatchers with complete situational awareness data containing accuracy confidence information.

[0083] In this embodiment of the invention, principal component analysis is used to decompose the state estimates of each partition, resulting in multiple signal data and multiple noise data. Random noise simulation is then performed based on the statistical characteristics of each noise data, yielding multiple simulated noise data. The sum between each signal data and its corresponding simulated noise data is calculated, resulting in multiple simulated measurement data. Orthogonal least squares is used to fit each simulated measurement data, yielding multiple fitting parameters. Based on a preset number of random simulations, the fitting parameters corresponding to the state estimates of each partition are cyclically sampled and statistically analyzed to obtain the corresponding distribution network state estimate.

[0084] It is worth mentioning that by correcting the state estimates of each partition, the multiple error defects caused by independent partition calculations can be specifically compensated. On the one hand, the correction process introduces an improved recursive least squares algorithm with bias compensation, which can compensate for the measurement disturbances caused by new energy fluctuations. This solves the estimation lag and parameter identification inaccuracy problems that occur when the traditional fixed forgetting factor algorithm encounters sudden changes in the power grid state. At the same time, it combines Monte Carlo simulation noise to statistically correct random measurement errors, weakening the impact of sensor noise and transmission interference on the state results and significantly improving the calculation accuracy of core state quantities such as voltage and power flow. On the other hand, the correction process completes the collaborative equalization calibration of the state of the boundary nodes of adjacent sub-regions, eliminating the problems of boundary state numerical mismatch and non-conservation of power across the entire network caused by partition decoupling calculations, which violate the physical constraints of the power grid. This ensures that the final integrated state estimate of the entire distribution network meets the operational constraints such as power balance, voltage limits, and line capacity.

[0085] In this embodiment of the invention, the power grid topology of the distribution network is acquired, and the topology is partitioned to obtain multiple sub-regions. Measurement data for each sub-region is acquired, and state evaluation is performed on each measurement data based on a preset least squares objective function to obtain multiple initial state data. Basic state prediction is then performed based on each measurement data and each initial state data to obtain multiple partition state estimates. State correction is then applied to each partition state estimate to obtain the corresponding distribution network state estimate. This overcomes the technical problem that existing distribution network state estimation methods often employ recursive least squares algorithms, which have a fixed forgetting factor, cannot adapt to sudden changes in new energy output, and are prone to estimation lag and large errors, thus reducing the accuracy of distribution network state estimation. Compared with traditional distribution network state estimation methods, this invention first decomposes a large-scale distribution network into multiple small sub-regions, significantly reducing the matrix operation dimension of a single state solution and effectively reducing the computing power consumption of a single node, thus meeting the timeliness requirements of real-time estimation for distributed distribution networks. Second, based on local measurement data of each region, the initial state is independently solved through a least-squares objective function, making full use of the redundancy of local measurements in each region and avoiding the measurement timing mismatch problem caused by global data transmission delay. The initial state results can quickly complete local coarse estimation, providing a reliable benchmark for time series prediction. Basic state prediction is performed based on each measurement data and each initial state data, breaking through the shortcomings of traditional static state estimation that relies only on single-moment measurements and cannot track the fluctuations of new energy sources and loads. It relies on the continuous characteristics of time series to capture the dynamic changes of the power grid caused by sudden changes in photovoltaic and wind power output, and outputs a regional state estimate that conforms to the time series evolution law, solving the problems of state lag and poor change tracking ability of traditional algorithms. Finally, state correction is performed on all regional state estimates to obtain an accurate distribution network state estimate, thereby improving the accuracy of distribution network state estimation.

[0086] Please see Figure 2 , Figure 2 This is a flowchart illustrating the steps of a power distribution network state estimation method provided in Embodiment 2 of the present invention.

[0087] This invention provides a method for estimating the state of a distribution network, comprising:

[0088] Step 201: Obtain the power grid topology of the distribution network, and partition the power grid topology to obtain multiple sub-regions.

[0089] In this embodiment of the invention, the power grid topology of the distribution network is obtained, the feeder division, node branch connection relationship, distributed power source access point and the deployment location of each acquisition terminal are sorted out, and the topology partitioning is carried out with the feeder as the basic unit. During the partitioning process, the observability verification of each block is completed simultaneously, the block boundary is adjusted to ensure that the measurement data of each block can independently support the state solution, and the common bus connecting the blocks is marked as the boundary node. Strongly coupled branches across regions are eliminated to avoid the internal calculation of the partition being affected by external fluctuations. After boundary fine-tuning and computing power balancing matching, the partition is solidified, resulting in multiple decoupled sub-regions that are only associated through the boundary nodes.

[0090] Step 202: Obtain measurement data for each sub-region, and perform state evaluation on each measurement data based on the preset least squares objective function to obtain multiple initial state data.

[0091] Further, step 202 includes the following sub-steps:

[0092] S11. Construct corresponding measurement vectors using each measurement data.

[0093] A measurement vector is a column vector that arranges the output values ​​of all measurement devices in a sub-region at the same time according to a preset fixed dimension order. Each row of elements corresponds to a specific measurement value, and the length of the vector is equal to the total number of measurement data in the sub-region.

[0094] In this embodiment of the invention, for any sub-region, its measurement data includes node measurement information such as node voltage amplitude and node injected power, as well as branch measurement information such as branch active power and branch reactive power. These heterogeneous measurement values ​​are sequentially filled into the corresponding row positions of the column vector according to a unified sorting rule to obtain multiple measurement vectors.

[0095] It should be noted that the sorting rule is usually to first group by measurement type, and then sort by node or branch number.

[0096] S12. Iteratively solve the preset least squares objective function using each measurement vector to obtain multiple initial state data.

[0097] Initial state data refers to the preliminary estimates of the node voltage magnitude and phase angle obtained by iteratively solving the least squares objective function in the first round of static estimation for each sub-region. These results are static cross-sectional estimates obtained solely based on the measurement data at the current single moment.

[0098] In this embodiment of the invention, a preset least-squares objective function is iteratively solved using various measurement vectors to obtain multiple initial state data. For example, firstly, a set of reasonable initial guess values ​​are assigned to the state variables of each sub-region. Then, these guess values ​​are substituted into the measurement function to calculate the theoretical measurement value and the difference between this value and the actual measurement vector is used to obtain the residual vector, which is then used to calculate the current value of the objective function. Subsequently, numerical optimization algorithms such as the Newton-Raphson method or orthogonal transformation method are used to linearize the objective function using the Jacobian matrix of the measurement function. By repeatedly calculating the correction direction and updating the estimated value of the state variables, the objective function value gradually decreases in each iteration until the change in the objective function value between two adjacent iterations is less than a preset convergence threshold or the number of iterations reaches the upper limit, at which point the objective function reaches its minimum point, and the corresponding state variable value is the initial state data of that sub-region.

[0099] It should be noted that the initial guess value is usually taken as the historical state value of the previous scheduling cycle or the flat start value (i.e., the voltage amplitude per unit value is set to 1.0 and the voltage phase angle is set to 0).

[0100] It should be noted that the Newton-Raphson method is a classic numerical optimization algorithm for solving nonlinear least squares problems. Its core idea is to perform a Taylor expansion of the nonlinear measurement function at the current estimation point in each iteration and retain the first-order linear term, thereby approximating the original nonlinear optimization problem into a linear least squares subproblem. By repeatedly solving the correction equation and updating the state variables, the objective function value is made to approach the minimum point successively. It has a second-order convergence speed and is the most commonly used iterative solution algorithm for power system state estimation.

[0101] It should be noted that the measurement function is as follows:

[0102] ;

[0103] ;

[0104] ;

[0105] ;

[0106] in, Let i be the active power of the branch flowing from node i to node j. Let be the voltage amplitude at node i, in per-unit values. Let be the voltage amplitude at node j, in per-unit values. Conductance of a branch circuit (line or transformer), measured in Siemens units (S). The susceptance of a branch circuit (line or transformer), measured in Siemens units. The voltage phase angle difference between node i and node j Let i be the reactive power flowing from node i to node j. Let be the total active power injected into node i. Let be the real part of the element in the i-th row and j-th column of the power system admittance matrix. Let be the imaginary part of the element in the i-th row and j-th column of the power system admittance matrix, i.e., mutual susceptance. Let j be the total reactive power injected into node i (i.e., the algebraic sum of power generation injection or load consumption), j be the first index of the node, and i be the second index of the node. This is the line-to-ground susceptance (i.e., half of the line's charging capacitance).

[0107] Step 203: Perform basic state prediction based on each measurement data and each initial state data to obtain multiple partition state estimates.

[0108] Furthermore, step 203 includes the following sub-steps:

[0109] S21. Construct the corresponding output measurement sequence and input measurement sequence using each measurement data.

[0110] The output measurement sequence refers to a one-dimensional time series vector formed by arranging the observation values ​​of a specific electrical quantity that characterizes the dynamic response of the system in the same sub-region at multiple consecutive sampling times in chronological order of sampling time. Typically, node voltage amplitude, node voltage phase angle, or power of key branches are selected as output quantities.

[0111] The input measurement sequence refers to a one-dimensional time series vector formed by arranging the observations of external excitation factors affecting the changes in system state within the same continuous time window in the same time sequence as the output measurement sequence. Typical external excitations include changes in the active power output and reactive power output of distributed power sources, and random power fluctuations of load nodes.

[0112] In this embodiment of the invention, corresponding output measurement sequences and input measurement sequences are constructed using various measurement data. The output measurement sequence is a time series vector formed by arranging the observations of electrical quantities (such as node voltage amplitude, node voltage phase angle, or branch power) characterizing the system response in the same sub-region at multiple consecutive sampling times in chronological order. The input measurement sequence is a time series vector formed by arranging the observations of external excitation factors affecting system state changes (such as active power output fluctuations of distributed power sources, reactive power output changes, and random power fluctuations of load nodes) within the same continuous time window in the same temporal order.

[0113] S22. Input the output measurement sequence and input measurement sequence corresponding to each measurement data into the preset first least squares estimation function to obtain multiple parameter vectors.

[0114] The parameter vector refers to a column vector composed of various coefficients that describe the dynamic characteristics of the system. Specifically, it includes the coefficients of each order of the autoregressive part (reflecting the degree of inertial influence of historical output at different lag times on the current output) and the coefficients of each order of the external input part (reflecting the degree of driving influence of historical input at different lag times on the current output).

[0115] In this embodiment of the invention, the output measurement sequence and input measurement sequence corresponding to each measurement data are respectively input into a preset first least squares estimation function. The first least squares estimation function is solved by minimizing the sum of squared errors between the theoretical output value of the model and the actual historical output value, and multiple parameter vectors are obtained.

[0116] It should be noted that the solution process is as follows: First, the input measurement sequence is arranged into an information matrix according to the format of the autoregressive moving average model. Each row of the matrix corresponds to a sampling time, and the row elements are composed of several historical input values ​​and historical output values ​​before that time according to a preset order. Then, the output measurement sequence is substituted into the objective function as the target vector. The linear least squares problem is solved by solving the normal equation or by using numerical methods such as orthogonal transformation, so as to obtain a set of parameter vectors that can best fit the historical input and output data of the sub-region.

[0117] It should be noted that the first least squares estimation function is specifically as follows:

[0118] ;

[0119] ;

[0120] ;

[0121] in, Let be the information vector at time t. The output quantity measurement at time t-1 is the actual observed value of the system response electrical quantity (such as node voltage amplitude, branch power, or PCC point frequency deviation) at the previous sampling time. For t- The output measurement value at time t, i.e., the current time. The output observation value at the previous sampling time, The input measurement value at time t-1 is the actual observed value of external excitation factors (such as changes in distributed power output, load power fluctuations, etc.) at the previous sampling time. For t- The input measurement value at time t, i.e., the value at time t before the current time. The input observations at each sampling time, This is the output order (autoregressive order). For the input order, Let be the parameter vector at time t. Let be the first autoregressive coefficient at time t. Let be the second autoregressive coefficient at time t. The first time at time t Autoregressive coefficient, Let be the first external input coefficient at time t. The second external input coefficient at time t. The first time at time t External input coefficients, Let be the parameter vector at time t. Forgetting factor, Let i be the information vector at time i. This is the output measurement value at time i.

[0122] S23. The preset second least squares estimation function is iteratively solved using the parameter vectors and initial state data corresponding to each measurement data to obtain multiple partition state estimates.

[0123] Partition state estimation refers to the dynamic optimal estimation result obtained by iteratively applying the second least squares estimation function to each sub-region at each time step. The result takes the node voltage magnitude and phase angle as the core output form and forms a continuous estimation trajectory in the time dimension. It contains both the accurate state information of the current section and the ability to predict future trends.

[0124] In this embodiment of the invention, the preset second least squares estimation function is iteratively solved using the parameter vectors and initial state data corresponding to each measurement data. By using the system dynamic relationship represented by the parameter vector as the basis for state transition, the state estimate value of the previous time moment is recursively pushed to the current time moment to obtain the state prediction value. Then, the prediction value is corrected and updated using the actual measurement data of the current time moment to obtain multiple partition state estimates.

[0125] It should be noted that the second least squares estimation function is specifically as follows:

[0126] ;

[0127] ;

[0128] ;

[0129] ;

[0130] in, For the partition state estimation at time t, These are the estimated values ​​of the basic parameters at time t. For time indexing, Let be the covariance matrix at time t. This is the deviation compensation coefficient. These are the estimated values ​​of the basic parameters at time t-1. Let be the corrected gain vector at time t. The output measurement value at time t. Let be the information vector at time t. Let be the covariance matrix at time t-1. Forgetting factor, It is an identity matrix.

[0131] Step 204: Use principal component analysis to decompose the state estimates of each partition to obtain multiple signal data and multiple noise data.

[0132] Principal component analysis (PCA) is a multivariate statistical dimensionality reduction technique based on the covariance structure of data. It projects the original high-dimensional data onto a set of mutually orthogonal principal component directions through linear transformation. Each principal component is arranged from largest to smallest according to its contribution rate to the total variance of the data it explains. This decomposes the total data variation into a signal-dominant part and a noise-dominant part, thereby achieving the separation of effective information from random error.

[0133] Signal data refers to the data matrix reconstructed by multiplying the principal components with the largest variance contribution rates with their corresponding load vectors. It represents the effective electrical state information contained in the original partition state estimation and is the most stable and representative structured component of the data.

[0134] Noisy data refers to the data matrix reconstructed by multiplying the principal components with small residual variance contribution rates with their corresponding load vectors. It represents the random measurement errors, model residuals, and unstructured interference components mixed in the original partition state estimation.

[0135] In this embodiment of the invention, principal component analysis is used to decompose the state estimates of each sub-region, resulting in multiple signal data and multiple noise data. For example, the state estimate data of each sub-region is organized into a two-dimensional data matrix. The rows of the matrix correspond to different sampling times or different dimensions of state variables, and the columns correspond to different measurement points or different types of electrical parameters. Eigenvalue decomposition of the covariance matrix is ​​performed on this data matrix to obtain a series of principal components and their corresponding eigenvectors arranged from largest to smallest eigenvalues. The first few principal components with larger eigenvalues ​​dominate the total variance, representing the most important change patterns and the most representative structural information in the data. The data reconstructed by multiplying these principal components with their corresponding load vectors is the signal data. The remaining principal components with smaller eigenvalues ​​explain a very small proportion of the variance and usually reflect unstructured interference components such as random fluctuations, measurement noise, and model residuals in the data. The data reconstructed from these remaining principal components is the noise data.

[0136] It should be noted that the specific expression for eigenvalue decomposition is as follows:

[0137] ;

[0138] in, The partition state estimation data is organized into a two-dimensional data matrix. The factor score matrix, For the factor loading matrix, The score vector of the k-th principal component, i.e., the matrix. The kth column, The loading vector of the k-th principal component, i.e., the matrix The kth row, For signal data, For noisy data, d is the number of principal factors, m is the second index of position, and k is the first index of position.

[0139] Step 205: Perform random noise simulation operations on each noise data to obtain multiple simulated noise data.

[0140] Random noise simulation refers to the technical operation of generating a large number of artificial noise samples that follow the same probability distribution after obtaining the statistical distribution characteristics of noise data, using computer random number generation algorithms. Its purpose is to realize scenarios by artificially constructing a variety of possible errors.

[0141] In this embodiment of the invention, the noise data corresponding to each sub-region is statistically analyzed to calculate key statistics such as mean and variance, and the probability distribution type is determined according to the data distribution characteristics. Based on the data distribution characteristics and probability distribution type, the corresponding simulated noise data is generated using a Monte Carlo random number generator.

[0142] It should be noted that the expression for the variance of noisy data is as follows:

[0143] ;

[0144] in, The total number of samples, The element in the i1-th row and j1-th column of the matrix E constructed for the noisy data. The total number of variables, The number of primary factors, i1 is the third index of position, and j1 is the fourth index of position.

[0145] Step 206: Add each signal data to its corresponding analog noise data to obtain multiple analog measurement data.

[0146] Analog measurement data refers to artificially synthesized measurement datasets obtained by adding signal data with a set of analog noise data. Mathematically, it simulates the measurement output of a real measuring device under conditions containing specific random errors. Each set of analog measurement data represents the measurement performance under a possible random error scenario.

[0147] In this embodiment of the invention, each signal data is algebraically added element by element to the corresponding analog noise data to obtain multiple analog measurement data.

[0148] Step 207: Perform orthogonal least squares fitting on each simulated measurement data to obtain multiple fitting parameters.

[0149] Orthogonal least squares fitting is an improved regression technique that introduces orthogonalization preprocessing on the basis of traditional least squares regression. By performing orthogonal decomposition (such as QR decomposition or singular value decomposition) on the independent variable matrix, the original independent variable space is transformed into a set of mutually orthogonal basis vectors, eliminating the multicollinearity effect among independent variables. Then, the parameter estimates are solved by substituting back the trigonometric equation system, so that the fitting process has better numerical stability and anti-interference ability.

[0150] Fitting parameters refer to the model coefficient vector obtained by substituting simulated measurement data into the regression model and performing orthogonal least squares fitting. Its structure is completely consistent with the parameter vector.

[0151] In this embodiment of the invention, orthogonal least squares fitting is performed on each simulated measurement data to obtain multiple fitting parameters. For example, each simulated measurement data is organized into an independent variable matrix and a dependent variable vector according to the format of the system dynamic model. The independent variable matrix is ​​composed of historical input and output values ​​at each time step combined according to a preset order, and the dependent variable vector is the output value measured at the current time step. Then, the independent variable matrix is ​​orthogonally decomposed into the product of an orthogonal matrix and an upper triangular matrix. The original linear least squares problem is transformed into a problem of solving a system of trigonometric equations by utilizing the properties of orthogonal matrices. The fitting parameters corresponding to the simulated measurement data can be obtained by back-substitution calculation.

[0152] Step 208: Based on a preset number of random simulations, perform cyclic sampling and statistical analysis on the fitting parameters corresponding to the state estimation of each partition to obtain the corresponding power distribution network state estimation.

[0153] Furthermore, step 208 includes the following sub-steps:

[0154] S31. When the number of fitting parameters is greater than or equal to the preset number of random simulations, the fitting parameters corresponding to the state estimation of each partition are averaged to obtain multiple fitting average values.

[0155] The number of fitting iterations refers to the number of orthogonal least squares fitting iterations that have been completed.

[0156] The number of random simulations refers to the pre-set total number of Monte Carlo simulation rounds.

[0157] In this embodiment of the invention, when the number of fitting parameters is greater than or equal to the preset number of random simulations, the mean value between the fitting parameters corresponding to the state estimation of each partition is calculated to obtain multiple fitting average values.

[0158] S32. Calculate the standard deviation of the fitting parameters corresponding to the state estimation of each partition, and obtain multiple fitting standard deviations.

[0159] In this embodiment of the invention, the standard deviation of the fitting parameters corresponding to the state estimation of each partition is determined as the fitting standard deviation.

[0160] S33. Use the average value and standard deviation of each fitted data as the corresponding distribution network state estimate.

[0161] In this embodiment of the invention, the average value of each fit and the standard deviation of each fit are determined as the corresponding distribution network state estimate.

[0162] S34. When the number of fitting parameters is less than the number of random simulations, then proceed to the step of performing random noise simulation operations based on each noise data to obtain multiple simulated noise data.

[0163] In this embodiment of the invention, when the number of fitting parameters is less than the number of random simulations, it indicates that the current size of the fitting parameters is still insufficient to support subsequent analysis with statistical significance. It is necessary to continue to expand the fitting parameters and jump to the step of performing random noise simulation operations based on each noise data to obtain multiple simulated noise data.

[0164] It is worth mentioning that, see Figure 3 As shown, the state estimation method of this application was compared with the traditional distribution network state estimation method under the same test conditions in multiple scenario simulations. The test system adopted the IEEE 33-node system with high-penetration distributed power sources and the improved IEEE 118-node system, and fitted the wind power and photovoltaic output curves based on the actual wind speed and solar irradiance data of the dispatch day to simulate the random fluctuation characteristics of the distribution network in real operation; from Figure 3 The results clearly show that during the relatively stable phase of system operation, the method of this invention maintains a low level of estimation error for node voltage amplitude and phase angle at each sampling time. The error curve is flat and without obvious oscillation. In contrast, the estimation error of the traditional method is generally higher and fluctuates more. This is mainly because the traditional method uses a fixed forgetting factor and cannot fully exploit the smoothing effect of historical data during the stable phase, making the estimation results susceptible to measurement noise interference.

[0165] In this embodiment of the invention, the power grid topology of the distribution network is acquired, and the topology is partitioned to obtain multiple sub-regions. Measurement data for each sub-region is acquired, and state evaluation is performed on each measurement data based on a preset least squares objective function to obtain multiple initial state data. Basic state prediction is then performed based on each measurement data and each initial state data to obtain multiple partition state estimates. State correction is then applied to each partition state estimate to obtain the corresponding distribution network state estimate. This overcomes the technical problem that existing distribution network state estimation methods often employ recursive least squares algorithms, which have a fixed forgetting factor, cannot adapt to sudden changes in new energy output, and are prone to estimation lag and large errors, thus reducing the accuracy of distribution network state estimation. Compared with traditional distribution network state estimation methods, this invention first decomposes a large-scale distribution network into multiple small sub-regions, significantly reducing the matrix operation dimension of a single state solution and effectively reducing the computing power consumption of a single node, thus meeting the timeliness requirements of real-time estimation for distributed distribution networks. Second, based on local measurement data of each region, the initial state is independently solved through a least-squares objective function, making full use of the redundancy of local measurements in each region and avoiding the measurement timing mismatch problem caused by global data transmission delay. The initial state results can quickly complete local coarse estimation, providing a reliable benchmark for time series prediction. Basic state prediction is performed based on each measurement data and each initial state data, breaking through the shortcomings of traditional static state estimation that relies only on single-moment measurements and cannot track the fluctuations of new energy sources and loads. It relies on the continuous characteristics of time series to capture the dynamic changes of the power grid caused by sudden changes in photovoltaic and wind power output, and outputs a regional state estimate that conforms to the time series evolution law, solving the problems of state lag and poor change tracking ability of traditional algorithms. Finally, state correction is performed on all regional state estimates to obtain an accurate distribution network state estimate, thereby improving the accuracy of distribution network state estimation.

[0166] Please see Figure 4 , Figure 4 This is a structural block diagram of a power distribution network state estimation system provided in Embodiment 3 of the present invention.

[0167] This invention provides a power distribution network state estimation system, comprising:

[0168] The acquisition module 301 is used to acquire the power grid topology of the distribution network, partition the power grid topology, and obtain multiple sub-regions.

[0169] The evaluation module 302 is used to acquire measurement data of each sub-region, and to evaluate the state of each measurement data based on a preset least squares objective function to obtain multiple initial state data.

[0170] Prediction module 303 is used to perform basic state prediction based on each measurement data and each initial state data to obtain multiple partition state estimates;

[0171] The correction module 304 is used to correct the state estimates of each zone to obtain the corresponding distribution network state estimates.

[0172] Furthermore, the evaluation module 302 includes:

[0173] The first construction submodule is used to construct the corresponding measurement vectors using each measurement data.

[0174] The first analysis submodule is used to iteratively solve the preset least squares objective function using each measurement vector to obtain multiple initial state data.

[0175] Furthermore, the prediction module 303 includes:

[0176] The second construction submodule is used to construct the corresponding output measurement sequence and input measurement sequence using each measurement data;

[0177] The second analysis submodule is used to input the output measurement sequence and input measurement sequence corresponding to each measurement data into a preset first least squares estimation function to obtain multiple parameter vectors.

[0178] The preset second least squares estimation function is iteratively solved using the parameter vectors corresponding to each measurement data and the initial state data to obtain multiple partition state estimates.

[0179] Furthermore, the correction module 304 includes:

[0180] The decomposition submodule is used to decompose the state estimates of each partition using principal component analysis to obtain multiple signal data and multiple noise data.

[0181] The simulation submodule is used to perform random noise simulation operations based on each noise data to obtain multiple simulated noise data.

[0182] The third analysis submodule is used to sum each signal data with the corresponding analog noise data to obtain multiple analog measurement data.

[0183] Orthogonal least squares fitting was performed on each of the simulated measurement data to obtain multiple fitting parameters;

[0184] The cyclic sampling statistics submodule is used to perform cyclic sampling statistics on the fitting parameters corresponding to the state estimates of each partition based on a preset number of random simulations, so as to obtain the corresponding power distribution network state estimates.

[0185] Furthermore, the cyclic sampling statistics submodule includes:

[0186] The mean unit is used to perform mean processing on the fitting parameters corresponding to the state estimates of each partition when the number of fitting parameters is greater than or equal to the preset number of random simulations, so as to obtain multiple fitting average values.

[0187] The bias unit is used to calculate the standard deviation of the fitting parameters corresponding to the state estimates of each partition, and obtain multiple fitting standard deviations.

[0188] The statistical unit is used to take the average value and standard deviation of each fit as the corresponding distribution network state estimate.

[0189] The jump unit is used to jump to the step of performing random noise simulation operation based on each noise data when the number of fitting parameters is less than the number of random simulations, so as to obtain multiple simulated noise data.

[0190] Furthermore, the second least squares estimation function is specifically as follows:

[0191] ;

[0192] ;

[0193] ;

[0194] ;

[0195] in, For the partition state estimation at time t, These are the estimated values ​​of the basic parameters at time t. For time indexing, Let be the covariance matrix at time t. This is the deviation compensation coefficient. These are the estimated values ​​of the basic parameters at time t-1. Let be the corrected gain vector at time t. The output measurement value at time t. Let be the information vector at time t. Let be the covariance matrix at time t-1. Forgetting factor, It is an identity matrix.

[0196] Please see Figure 5 , Figure 5 This is a structural block diagram of an electronic device provided in Embodiment 4 of the present invention.

[0197] An electronic device according to an embodiment of the present invention includes: a memory 401 and a processor 402. The memory 401 stores a computer program. When the computer program is executed by the processor 402, the processor 402 performs a power distribution network state estimation method as described in any of the above embodiments.

[0198] Memory 401 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 401 has storage space 403 for program code 413 for performing any of the method steps described above. For example, storage space 403 for program code may include individual program codes 413 for implementing the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. This program code may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When this code is run by a computing processing device, it causes the device to perform the various steps in the distribution network state estimation method described above.

[0199] Embodiment 5 of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the power distribution network state estimation method as described in any of the above embodiments.

[0200] Embodiment 6 of the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the power distribution network state estimation method as described in any of the above embodiments.

[0201] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0202] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0203] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0204] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0205] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0206] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating the state of a distribution network, characterized in that, include: Obtain the power grid topology of the distribution network, and partition the power grid topology to obtain multiple sub-regions; Acquire measurement data for each of the sub-regions, and perform state evaluation on each of the measurement data based on a preset least squares objective function to obtain multiple initial state data. Based on each of the measurement data and the initial state data, basic state prediction is performed to obtain multiple partition state estimates. The state estimates of each partition are corrected to obtain the corresponding distribution network state estimates.

2. The distribution network state estimation method according to claim 1, characterized in that, The step of performing state evaluation on each of the measurement data based on a preset least squares objective function to obtain multiple initial state data includes: Each measurement data point is used to construct a corresponding measurement vector. The preset least squares objective function is solved iteratively using each of the aforementioned measurement vectors to obtain multiple initial state data.

3. The distribution network state estimation method according to claim 1, characterized in that, The step of performing basic state prediction based on each of the measurement data and each of the initial state data to obtain multiple partition state estimates includes: The corresponding output measurement sequence and input measurement sequence are constructed using each of the aforementioned measurement data; The output measurement sequence and input measurement sequence corresponding to each of the measurement data are respectively input into a preset first least squares estimation function to obtain multiple parameter vectors; The preset second least squares estimation function is iteratively solved using the parameter vectors and initial state data corresponding to each of the measurement data to obtain multiple partition state estimates.

4. The distribution network state estimation method according to claim 1, characterized in that, The step of performing state correction on each of the partition state estimates to obtain the corresponding distribution network state estimate includes: Principal component analysis was used to decompose the state estimates of each partition, resulting in multiple signal data and multiple noise data. Random noise simulation operations are performed on each of the aforementioned noise data to obtain multiple simulated noise data. Each of the aforementioned signal data and its corresponding analog noise data is summed to obtain multiple analog measurement data. Orthogonal least squares fitting was performed on each of the simulated measurement data to obtain multiple fitting parameters; Based on a preset number of random simulations, the fitting parameters corresponding to the state estimates of each partition are cyclically sampled and statistically analyzed to obtain the corresponding power distribution network state estimates.

5. The distribution network state estimation method according to claim 4, characterized in that, The step of cyclically sampling and statistically analyzing the fitting parameters corresponding to the state estimates of each partition based on a preset number of random simulations to obtain the corresponding distribution network state estimates includes: When the number of fitting parameters is greater than or equal to the preset number of random simulations, the fitting parameters corresponding to each partition state estimate are averaged to obtain multiple fitting average values. Calculate the standard deviation of the fitting parameters corresponding to each partition state estimate to obtain multiple fitting standard deviations; The average value and standard deviation of each of the aforementioned fitting values ​​are used as the corresponding power distribution network state estimates; When the number of fitting attempts for the fitting parameters is less than the number of random simulations, the process jumps to the step of performing random noise simulation operations based on each of the noise data to obtain multiple simulated noise data.

6. The distribution network state estimation method according to claim 3, characterized in that, The second least squares estimation function is as follows: ； ； ； ； in, For the partition state estimation at time t, These are the estimated values ​​of the basic parameters at time t. For time indexing, Let be the covariance matrix at time t. This is the deviation compensation coefficient. These are the estimated values ​​of the basic parameters at time t-1. Let be the corrected gain vector at time t. The output measurement value at time t. Let be the information vector at time t. Let be the covariance matrix at time t-1. Forgetting factor, It is an identity matrix.

7. A power distribution network state estimation system, characterized in that, include: The acquisition module is used to obtain the power grid topology of the distribution network, and to partition the power grid topology to obtain multiple sub-regions; The evaluation module is used to acquire measurement data of each of the sub-regions, and to evaluate the state of each measurement data based on a preset least squares objective function to obtain multiple initial state data. The prediction module is used to perform basic state prediction based on each of the measurement data and each of the initial state data to obtain multiple partition state estimates. The correction module is used to correct the state estimates of each partition to obtain the corresponding distribution network state estimates.

8. An electronic device, characterized in that, The system includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the power distribution network state estimation method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the power distribution network state estimation method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the power distribution network state estimation method as described in any one of claims 1-6.

Images