Power distribution network state estimation method and device based on high-order volume Kalman filtering

Abstract

The invention discloses a power distribution network state estimation method and device based on high-order volume Kalman filtering, and the method comprises the steps: firstly calculating a volume point of a state variable of a power distribution network at a current moment, calculating a propagation volume point according to a state equation of the power distribution network, and calculating the state variable of the power distribution network according to the propagation volume point; and calculating a state predicted value at the next moment and a predicted value of the state error covariance matrix, and performing volume transformation on the propagation volume point to obtain a volume point at the next moment. Then, propagation is carried out on a volume point at the next moment through a measurement function, and a measurement predicted value at the next moment, a measurement error covariance matrix and a cross covariance matrix are calculated; and finally, calculating a Kalman filtering gain according to the measurement error covariance matrix and the cross covariance matrix, correcting a state prediction value by adopting the calculated Kalman filtering gain, and updating the state error covariance matrix, so that the state estimation precision is improved, and the estimation accuracy is improved. And the robustness and the precision of the system facing bad data are effectively enhanced.

Description

Technical Field

[0001] This application belongs to the field of distribution network state estimation technology, and particularly relates to a distribution network state estimation method and apparatus based on high-order capacitive Kalman filtering. Background Technology

[0002] The distribution network is a crucial component of the power system, responsible for distributing and transmitting electricity from the high-voltage grid to various sectors, including homes and businesses. Real-time dynamic status information is essential for the monitoring and safety control of the distribution network. However, with the continuous integration of unstable and intermittent energy sources such as wind and solar power into the grid, the electrical quantities of the distribution network are prone to transient changes. Simultaneously, the diverse load composition and complex line topology of modern distribution networks pose significant challenges to the accuracy of dynamic state estimation.

[0003] Generally, existing literature identifies two basic state estimation methods: Static State Estimation (SSE) and Dynamic State Estimation (DSE). Static estimation algorithms based on the least squares method can handle the state estimation problem of distribution networks over a specific time period. Synchronous Phasor Measurement Units (PMUs) and Supervisory Control and Data Acquisition (SCADA) systems are used to improve the accuracy of state estimation. SSE, considering the distribution network topology, calculates the distribution state using measurements at a given moment. However, it cannot obtain the system state at the next moment. Therefore, each state estimation operation requires re-initialization, which incurs significant time and computational costs. Furthermore, SSE does not consider the latest measurement results during execution, causing the estimated result to always lag behind the actual system state. With the widespread adoption of PMUs and advanced measurement equipment, the amount of data that can be collected from distribution systems has greatly increased. In this context, DSE for distribution networks is proposed. Combining the predicted state value from the previous moment with the current measurement value, DSE obtains the distribution network state, meeting the real-time requirements of distribution network state estimation.

[0004] One of the most commonly used distribution network state estimation (DSE) methods is based on the Extended Kalman Filter (EKF) algorithm, as well as two-step fault-tolerant EKF and adaptive interpolation strong tracking EKF developed based on EKF. However, EKF and its improved algorithms may introduce truncation errors due to the first-order Taylor expansion approximation. Due to the strong nonlinearity of the distribution network system equations, the accuracy of DSE is significantly reduced. To overcome the shortcomings of the EKF algorithm, researchers have proposed an unscented Kalman filter (UKF) based on unscented transform to perform DSE. Furthermore, considering the uncertainty of unknown noise and power grid data transmission, UKF-based distribution network state estimation has been extensively studied. Although UKF is effective in handling strongly nonlinear systems, its state estimation performance deteriorates with increasing state variable dimension. Therefore, UKF performs poorly when estimating large-scale distribution networks. In addition, a third-order precision capacitive Kalman filter (CKF) for power system state estimation has been proposed. To improve the robustness and accuracy of DSE in the presence of malicious information, a robust CKF based on random events and an adaptive CKF with generalized entropy loss have been proposed. However, with the integration of large-scale new energy sources and the complexity of power distribution line topologies, the estimation accuracy of the CKF-based DSE method is limited. Summary of the Invention

[0005] The purpose of this application is to provide a method and apparatus for estimating the state of a distribution network based on high-order capacitive Kalman filtering, so as to improve the reliability and accuracy of the estimation.

[0006] To achieve the above objectives, the technical solution of this application is as follows:

[0007] A distribution network state estimation method based on high-order capacitive Kalman filtering includes:

[0008] Step 1: Calculate the volume points of the distribution network state variables at the current moment, and calculate the propagation volume points according to the distribution network state equations. The number of volume points is... , where n is the dimension of the state variable;

[0009] Step 2: Based on the propagation volume point, calculate the predicted state value and the predicted state error covariance matrix for the next time step.

[0010] Step 3: Perform a volume transformation on the propagation volume point to obtain the volume point at the next moment;

[0011] Step 4: Propagate the volume point at the next time step using the measurement function, calculate the predicted measurement value at the next time step, as well as the measurement error covariance matrix and cross-covariance matrix;

[0012] Step 5: Calculate the Kalman filter gain based on the measurement error covariance matrix and the cross-covariance matrix. Then, use the calculated Kalman filter gain to correct the state prediction value and update the state error covariance matrix.

[0013] Preferably, after step 4, the method further includes:

[0014] Anomaly detection is performed on the measurement data. If the data is determined to be abnormal, the measurement error covariance matrix and cross-covariance matrix are adaptively corrected.

[0015] Preferably, the step of anomaly detection of the measurement data, and adaptive correction of the measurement error covariance matrix and cross-covariance matrix if the data is determined to be abnormal, includes:

[0016] Define the new information vector By judging the new information matrix traces The trace of the measurement error covariance matrix The size of the adaptive factor is used to construct the adaptive factor. :

[0017] ;

[0018] in, Indicates the actual measured value. Indicates the predicted measurement value. Represents the measurement error covariance matrix;

[0019] Calculate the Euclidean norm of the innovation vector. If the Euclidean norm of the innovation vector is within the range of norm values, then use the constructed adaptive factor to correct the measurement error covariance matrix and cross-covariance matrix, as shown in the following formula:

[0020] ;

[0021] in, For weights, This represents the set of points obtained by propagating the volume point at the next time step through the measurement function. This represents the volume point at the next moment obtained by performing a volume transformation on the propagation volume point. express The state prediction value at time t, where R represents the variance of the observed white noise.

[0022] Preferably, if the Euclidean norm of the innovation vector is greater than the maximum value of the norm value distribution, then the predicted state value is directly used as the estimated state vector value, and the predicted state error covariance matrix is ​​used as the estimated state error covariance matrix value.

[0023] Preferably, if the Euclidean norm of the innovation vector is less than the minimum value of the norm value distribution, then step 5 is executed directly.

[0024] Preferably, the step of calculating the Kalman filter gain based on the measurement error covariance matrix and the cross-covariance matrix, then using the calculated Kalman filter gain to correct the state prediction value and update the state error covariance matrix includes:

[0025] The Kalman filter gain is calculated using the following formula. :

[0026] ;

[0027] Then, combine the actual measured values and measurement prediction value For the predicted state value The state estimate is obtained by performing correction. :

[0028] ;

[0029] Meanwhile, the state error covariance matrix Update the formula as follows:

[0030] ;

[0031] in, This represents the predicted value of the state error covariance matrix. Represents the measurement error covariance matrix. This represents the cross-covariance matrix.

[0032] This application also proposes a distribution network state estimation device based on high-order capacitive Kalman filtering, including a processor and a memory storing a number of computer instructions, which, when executed by the processor, implement the steps of the above method.

[0033] This application proposes a distribution network state estimation method and apparatus based on high-order capacitive Kalman filtering (HCKF). The HCKF uses a novel estimator based on HCKF to study the state estimation method. HCKF uses the fifth-order capacitive rule to calculate the average probability and error covariance matrix, improving the accuracy of state estimation. It approximates the system's probability distribution by using increasingly refined capacitive points. Therefore, HCKF has significant advantages over CKF when dealing with strongly nonlinear and high-dimensional distributed networks. This application also reduces the impact of bad data on DSE by constructing adaptive factor pairs for online correction, effectively enhancing the system's robustness and accuracy in the face of bad data. Attached Figure Description

[0034] Figure 1 This is a flowchart of the power distribution network state estimation method based on high-order capacitive Kalman filtering proposed in this application. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0036] To facilitate the explanation of the method proposed in this application, this application first describes the dynamic state estimation model of the distribution network.

[0037] Because the daily load changes slowly and random fluctuations are small within the distribution area, the distribution network can be assumed to operate under quasi-static conditions. The distribution network state estimation model consists of system state equations and measurement equations, as shown below:

[0038] (1)

[0039] (2)

[0040] Among them, subscript Indicates time, Represents an n-dimensional real-valued state variable. and Let represent the voltage magnitude and phase angle of the i-th node, respectively; This represents an m-dimensional real-valued measurement vector, which includes node voltage magnitude and phase angle, node injected active power, node injected reactive power, branch active power, and branch reactive power. This represents a nonlinear function for state transition. This indicates a measurement nonlinear function; Represents the system process white noise. Let represent observed white noise, with all values ​​having zero mean and variances of . and , This represents the variance of white noise in the system process. This represents the variance of the observed white noise.

[0041] Since Holt's two-parameter exponential smoothing is a time series forecasting method, it is suitable for relatively stable time series data because it estimates future trends based on a weighted average of past observations. Using this method, the state transition function... Decomposed into horizontal components and smooth tilt components As shown below:

[0042] (3)

[0043] In the formula: It is the predicted state value from time k to time k+1; It is the state estimate at time k; For horizontal smoothing parameters, The smoothing parameter is the key to the accuracy of the distribution network model, and is usually set to [0,1]. After rearranging equation (3), the system state equation can be obtained:

[0044] (4)

[0045] in, and These represent the degree of dependence on short-term and long-term data, respectively. Since distribution network load exhibits a clear daily variation pattern, the optimal smoothing parameters can be obtained through historical daily load curves. In the absence of daily load data, measurement data of a sample distribution system can be obtained through Newton-Raphson power flow analysis. Then, measurements conforming to specific rules are selected to form a sample measurement set. Furthermore, zero-mean Gaussian measurement errors are injected into these measurements to obtain data similar to the actual measurements in the distribution system.

[0046] Due to economic and technological limitations, phasor measurement units (PMUs) cannot replace SCADA in the distribution network in the short term. To maximize the use of existing observation data, a hybrid approach of PMUs and SCADA is often used in distribution networks to provide measurements for dynamic state estimation. PMUs can measure node voltage and phase angle, while SCADA can measure node voltage magnitude, injected active power, injected reactive power, branch active power, and branch reactive power. The PMU measurement equation can be expressed as:

[0047] (5)

[0048] in, This indicates the PMU measurement noise, measurement function. for:

[0049] (6)

[0050] in, and Let represent the voltage magnitude and phase angle measured at the i-th node via the PMU, respectively. The SCADA measurement equation can be expressed as:

[0051] (7)

[0052] in, This indicates SCADA measurement noise.

[0053] make Ignoring node-to-ground conductance and susceptance, It can be represented as:

[0054] (8)

[0055] in, , Indicates the node label. Represents a node Injected active power, Represents a node and Active power of the branches between Indicates the voltage magnitude at node i. This indicates that reactive power is injected into node i. Represents a node and reactive power of the branches between; Represents a node , The voltage phase angle difference between them; , Representing nodes respectively and The conductance and susceptance of the branches between them; , These represent the first and second nodes of the nodal admittance matrix, respectively. Line number The real and imaginary parts of a column element; symbol In Represents all nodes Connected nodes.

[0056] Combining (5) and (7), the hybrid measurement model consisting of the PMU measurement equation and the SCADA measurement equation is as follows:

[0057] (9)

[0058] Numerous sensors are deployed in the power distribution system, making the entire system fully observable. These sensors provide measurements of active and reactive power at nodes and branches. Furthermore, nonlinear measurement equations can be derived based on Ohm's law and power distribution line parameters. Assuming slow load changes in the distribution network, the changes in the network state conform to a Holter-based dynamic model.

[0059] Example 1: Based on the above reasoning regarding the distribution network state estimation model, this application proposes a distribution network state estimation method based on high-order capacitive Kalman filtering, including:

[0060] Step S1: Calculate the volume points of the distribution network state variables at the current moment, and calculate the propagation volume points according to the distribution network state equations. The number of volume points is... , where n is the dimension of the state variable.

[0061] The main difference between the high-order capacitive Kalman filter (HCKF) and the third-order precision capacitive Kalman filter (CKF) in this embodiment lies in the selection of the volume points and the method of calculating the weights. HCKF selection... The number of volume points is n, where n is the dimension of the state variables. By increasing the number of volume points, the accuracy of the nonlinear function approximation model is improved. In this case, nonlinear changes in the distribution network state can be accurately captured. Therefore, theoretically, HCKF has a lower estimation bias. This application proposes a distribution network state estimation algorithm based on HCKF (DSE-HCKF).

[0062] This step calculates the current... Capacity point of distribution network state variables at any time :

[0063] (10)

[0064] In the formula: yes Estimates of state variables at time points; for Estimates of the time-state error covariance matrix; It is the dimension of the state variables. It is by The lower triangular matrix obtained by Cholesky decomposition, that is .make ,vector for:

[0065] (11)

[0066] in , It is 3D vector for dimensional unit vector and the th Behavior 1, point set and As shown respectively and .

[0067] The volume point is obtained by propagating the state equation of the nonlinear system to the volume point of equation (10). :

[0068] (12).

[0069] The distribution network state estimation algorithm based on HCKF in this application includes a state prediction process and a measurement update process. In the state prediction process, a selection... Increasing the number of volumetric points is an important step in improving estimation accuracy. Because the hybrid measurement equations for distribution networks are highly nonlinear, increasing the number of volumetric points can improve the accuracy of the measurement function approximation during measurement updates, enabling more accurate capture of nonlinear changes in the distribution network state.

[0070] Step S2: Calculate the predicted state value and the predicted state error covariance matrix for the next time step based on the propagation volume point.

[0071] This step calculates the propagation volume point based on equation (12). State prediction value at time 1 The predicted value of the state error covariance matrix :

[0072] (13)

[0073] Among them, weight The value can be:

[0074] (14).

[0075] in, express The predicted state value at time 10:00. This represents the predicted value of the state error covariance matrix. .

[0076] Step S3: Perform a volume transformation on the propagation volume point to obtain the volume point at the next moment.

[0077] The formula is as follows:

[0078] (15)

[0079] in .

[0080] Step S4: Propagate the volume point at the next time step through the measurement function, calculate the measurement prediction value at the next time step, as well as the measurement error covariance matrix and cross-covariance matrix.

[0081] This step first obtains the point set using the following formula. :

[0082] (16)

[0083] This represents the set of points obtained by propagating the volume point to the next time step through the measurement function; it has no actual physical meaning. Then, the calculation... Predicted measurement value at time:

[0084] (17)

[0085] calculate Measurement error covariance matrix at time and cross-covariance matrix :

[0086] (18)

[0087] in, for transpose, for Predicted measurement value at time. for transpose, This represents the volume point at the next moment obtained by performing a volume transformation on the propagation volume point.

[0088] Step S5: Calculate the Kalman filter gain based on the measurement error covariance matrix and the cross-covariance matrix. Then, use the calculated Kalman filter gain to correct the state prediction value and update the state error covariance matrix.

[0089] The Kalman filter gain is calculated using the following formula. :

[0090] (twenty two).

[0091] Then, combine the actual measured values and measurement prediction value The state vector estimate is obtained by correcting the state prediction value. :

[0092] (twenty three)

[0093] Simultaneously, the state error covariance matrix is ​​updated using the following formula:

[0094] (twenty four)

[0095] This is an estimate of the state error covariance matrix.

[0096] It should be noted that after obtaining the estimated state vector and the state error covariance matrix, the distribution network state at subsequent times can be estimated using the previous method, which will not be elaborated here.

[0097] Example 2: Due to factors such as measurement errors, equipment failures, malicious interference, and changes in the external environment, the observed data may be poor, leading to a degradation in state estimation performance. To improve the adaptability of the HCKF algorithm when poor data occurs, this example, based on Example 1, constructs an adaptive factor to improve DSE accuracy.

[0098] That is, after step S4 in Example 1, the following step was added:

[0099] Anomaly detection is performed on the measurement data. If the data is determined to be abnormal, the measurement error covariance matrix and cross-covariance matrix are adaptively corrected.

[0100] This embodiment reduces the impact of adverse data on DSE by constructing an adaptive factor to adaptively correct these two covariance matrices. The step of anomaly detection of measurement data, and adaptive correction of the measurement error covariance matrix and cross-covariance matrix if the data is determined to be abnormal, includes:

[0101] Step 1.1: Define the innovation vector By judging the new information matrix traces The trace of the measurement error covariance matrix The size of the adaptive factor is used to construct the adaptive factor. :

[0102] (19)

[0103] The trace is a mathematical operation on matrices, which will not be elaborated here.

[0104] Step 1.2: Calculate the Euclidean norm of the innovation vector. If the Euclidean norm of the innovation vector is within the range of norm values, then use the constructed adaptive factor to correct the measurement error covariance matrix and cross-covariance matrix.

[0105] To identify and evaluate bad data, the Euclidean norm was used as the criterion. The Euclidean norm describes the degree of dissimilarity between two samples by calculating the true distance between them; its expression is:

[0106] (20)

[0107] Here, i represents the i-th element in the vector, and m is the dimension of the observation. By calculating the Euclidean norm of the innovation vector at each time step, a norm value distribution can be obtained. The maximum and minimum values ​​of the norm are denoted as... and Then the norm value distribution is . This represents the i-th element in the measured actual value. This represents the i-th element in the measurement prediction value.

[0108] If the Euclidean norm of the innovation vector is in Within this matrix, the trace of the innovation matrix is ​​greater than the trace of the measurement prediction error variance matrix, indicating that there is bad data in the system measurements.

[0109] The measurement prediction variance needs to be corrected online using a constructed adaptive factor, as shown in the following formula:

[0110] (twenty one)

[0111] It should be noted that if the Euclidean norm of the innovation vector is greater than... At this point, even using the adaptive factor method described above, the DSE effect is not ideal. This is because the observed values ​​are abnormal and deviate too much from the true values. Therefore, the predicted state value can be used as the estimated state vector value, i.e. The predicted value of the state error covariance matrix is ​​used as the estimated value of the state error covariance matrix, i.e. That is to say, step S5 of embodiment 1 will no longer be executed.

[0112] When the Euclidean norm of the innovation vector is less than At this point, the system is operating normally without any abnormal values. Subsequently, step S5 in Example 1 is executed directly, calculating the Kalman filter gain based on the measurement error covariance matrix and the cross-covariance matrix. Then, the calculated Kalman filter gain is used to correct the state prediction value and update the state error covariance matrix.

[0113] It is easy to understand that the Kalman gain The design of the filter estimation directly affects the accuracy of the filter estimation, and it is mainly composed of the autocovariance matrix. With the cross-covariance matrix Therefore, the impact of adverse data on DSE is reduced by constructing an adaptive factor to correct the two covariance matrices online. This embodiment effectively handles adverse data in the distribution network by introducing the Euclidean norm and the adaptive factor. First, the algorithm uses the Euclidean norm to evaluate the difference between the measured data and normal data, thereby determining whether the data is abnormal. When the Euclidean norm exceeds a preset threshold, it indicates that the data has significant anomalies, and the system will adopt different processing strategies. For mildly adverse data, the system corrects the covariance matrix through the adaptive factor and adjusts the filter parameters to minimize the impact of adverse data on the estimation results; while for severely adverse data, continuing to use the observed data for estimation may introduce large errors. At this time, the system directly uses a one-step prediction as the state estimate, relying entirely on the prediction results of the system model. Since the model prediction is not affected by the abnormal data, it can provide a more accurate estimate than the observed data, thereby avoiding the expansion of estimation errors. This algorithm effectively enhances the robustness and accuracy of the system when facing adverse data through dynamic correction and online judgment.

[0114] Finally, to verify the effectiveness of the technical solution presented in this application, experiments were conducted using an IEEE 12-node distributed system and an IEEE 33-node distributed system. Furthermore, to better compare the estimation performance of each algorithm, the average and maximum values ​​of the relative errors were used as performance metrics.

[0115] First, the technical solution of Embodiment 1 of this application (DSE-HCKF) and the dynamic state estimation algorithm based on CKF (DSE-CKF) are compared to demonstrate the accuracy of the method proposed in this application.

[0116] The experimental statistics for the IEEE 12-node distributed system experiment are shown in Table 1.

[0117] Table 1

[0118]

[0119] As can be seen, the average and maximum relative errors of the DSE-HCKF algorithm are smaller than those of the DSE-CKF algorithm, especially the amplitude relative error. The average and maximum amplitude relative errors of DSE-CKF are more than twice as large as those of DSE-HCFK. Therefore, the above results demonstrate the accuracy of the proposed DSE-HCKF algorithm in the IEEE 12-node distributed system.

[0120] The experimental statistics for the IEEE 33-node distributed system experiment are shown in Table 2.

[0121] Table 2

[0122]

[0123] Table 2 shows the average and maximum values ​​of the voltage vector relative error for the two algorithms. We can see that DSE-HCKF has the smallest average and maximum relative error. Compared to DSE-CKF, DSE-HCKF reduces the average and maximum voltage relative error by 57.9% and 32.2%, respectively. The average phase angle relative error and maximum voltage relative error of DSE-HCKF are also lower than those of DSE-CKF by 28.9% and 19%, respectively. Therefore, it can be concluded that DSE-HCKF is more accurate than DSE-CKF.

[0124] Next, the technical solutions of Embodiment 2 (DSE-AHCKF) and Embodiment 1 (DSE-HCKF) of this application were compared in an IEEE 33-node distributed system. The experimental results are shown in Table 3.

[0125] Table 3

[0126]

[0127] Table 3 shows the average and maximum values ​​of the voltage vector relative errors for the two algorithms. It can be seen that the DSE-AHCKF algorithm has the smallest average and maximum relative errors. Specifically, the average and maximum voltage relative errors of the DSE-AHCKF algorithm are reduced by 66.7% and 91.5% respectively compared to the DSE-HCKF algorithm. The average phase angle and maximum voltage relative errors of the DSE-AHCKF algorithm are reduced by 37.6% and 92.7% respectively compared to the DSE-HCKF algorithm. Therefore, it can be concluded that the DSE-AHCKF algorithm significantly outperforms the DSE-HCKF algorithm in state estimation when dealing with poor data.

[0128] Example 3: This example also provides a distribution network state estimation device based on high-order capacitive Kalman filtering, including a processor and a memory storing a number of computer instructions. When the computer instructions are executed by the processor, they implement the steps of the above-mentioned distribution network state estimation method based on high-order capacitive Kalman filtering.

[0129] Specific limitations regarding the distribution network state estimation device based on high-order ductile Kalman filtering can be found in the limitations of the distribution network state estimation method based on high-order ductile Kalman filtering mentioned above, and will not be repeated here. The aforementioned distribution network state estimation device based on high-order ductile Kalman filtering can be implemented entirely or partially through software, hardware, or a combination thereof. It can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations described above.

[0130] The memory and processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines. The memory stores a computer program that can run on the processor, which implements the power distribution network state estimation method based on high-order capacitive Kalman filtering in this embodiment of the invention by running the computer program stored in the memory.

[0131] The memory may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory stores the program, and the processor executes the program upon receiving an execution instruction.

[0132] The processor may be an integrated circuit chip with data processing capabilities. The aforementioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor.

[0133] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A distribution network state estimation method based on high-order capacitive Kalman filtering, characterized in that, The distribution network state estimation method based on high-order capacitive Kalman filtering includes: Step 1: Calculate the volume points of the distribution network state variables at the current moment, and calculate the propagation volume points according to the distribution network state equations. The number of volume points is... , where n is the dimension of the state variable; Step 2: Based on the propagation volume point, calculate the predicted state value and the predicted state error covariance matrix for the next time step. Step 3: Perform a volume transformation on the propagation volume point to obtain the volume point at the next moment; Step 4: Propagate the volume point at the next time step using the measurement function, calculate the predicted measurement value at the next time step, as well as the measurement error covariance matrix and cross-covariance matrix; Step 5: Calculate the Kalman filter gain based on the measurement error covariance matrix and the cross-covariance matrix. Then, use the calculated Kalman filter gain to correct the state prediction value and update the state error covariance matrix.

2. The distribution network state estimation method based on high-order capacitive Kalman filtering according to claim 1, characterized in that, After step 4, the following is also included: Anomaly detection is performed on the measurement data. If the data is determined to be abnormal, the measurement error covariance matrix and cross-covariance matrix are adaptively corrected.

3. The distribution network state estimation method based on high-order capacitive Kalman filtering according to claim 2, characterized in that, The process of anomaly detection for measurement data, including adaptive correction of the measurement error covariance matrix and cross-covariance matrix if anomalies are detected, includes: Define the new information vector By judging the new information matrix traces The trace of the measurement error covariance matrix The size of the adaptive factor is used to construct the adaptive factor. : ； in, Indicates the actual measured value. Indicates the predicted measurement value. Represents the measurement error covariance matrix; Calculate the Euclidean norm of the innovation vector. If the Euclidean norm of the innovation vector is within the range of norm values, then use the constructed adaptive factor to correct the measurement error covariance matrix and cross-covariance matrix, as shown in the following formula: ； in, For weights, This represents the set of points obtained by propagating the volume point at the next time step through the measurement function. This represents the volume point at the next moment obtained by performing a volume transformation on the propagation volume point. express The state prediction value at time t, where R represents the variance of the observed white noise.

4. The distribution network state estimation method based on high-order capacitive Kalman filtering according to claim 3, characterized in that, If the Euclidean norm of the innovation vector is greater than the maximum value of the norm value distribution, then the predicted state value is directly used as the estimated state vector value, and the predicted state error covariance matrix is ​​used as the estimated state error covariance matrix value.

5. The distribution network state estimation method based on high-order capacitive Kalman filtering according to claim 3, characterized in that, If the Euclidean norm of the innovation vector is less than the minimum value of the norm value distribution, then proceed directly to step 5.

6. The distribution network state estimation method based on high-order capacitive Kalman filtering according to claim 1, characterized in that, The process of calculating the Kalman filter gain based on the measurement error covariance matrix and the cross-covariance matrix, then using the calculated Kalman filter gain to correct the state prediction value, and updating the state error covariance matrix includes: The Kalman filter gain is calculated using the following formula. : ； Then, combine the actual measured values and measurement prediction value For the predicted state value The state estimate is obtained by performing correction. : ； Meanwhile, the state error covariance matrix To update, the formula is: ； in, This represents the predicted value of the state error covariance matrix. Represents the measurement error covariance matrix. This represents the cross-covariance matrix.

7. A distribution network state estimation device based on high-order capacitive Kalman filtering, comprising a processor and a memory storing a plurality of computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 6.

Images