Unobservable power distribution system state estimation method based on random switching subsystem

Abstract

The invention relates to an unobservable power distribution system state estimation method based on a random switching subsystem, and the method comprises the steps: determining a state variable of state estimation through inputting a line impedance parameter, a load, the access position and capacity of a distributed power supply, and the configuration information of multi-source measurement data; calculating the rank of the observability matrix in the current time period according to the line impedance parameter and the configuration information of the multi-source measurement data; constructing a transfer matrix to convert the state variable; solving a converted state variable by adopting an unscented Kalman filtering method; and constructing an inverse transformation matrix to restore the converted state variable, and outputting a state estimation result. According to the method, the power distribution network is modeled into a random switching subsystem under an unobservable condition, a switching path of the power distribution network is disclosed, frequent switching of the system is reduced, a high-dimensional state variable is mapped to a low-dimensional conversion variable through state conversion, an improved unscented Kalman filtering algorithm is used for solving, and the precision and speed of state estimation are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid assessment, and in particular to an unobservable distribution system state estimation method based on a random switching subsystem. Background Art

[0002] With the high penetration rate of distributed power sources, the operating status of the distribution network urgently needs higher accuracy and faster tracking. State estimation can obtain the operating status of the entire network based on the measurement data under specific parameters, and is a basic tool for real-time monitoring, scheduling and regulation of the distribution network. However, the actual distribution network points are numerous and wide, and it is difficult to fully cover the measurement, resulting in a large number of "blind spots". A major challenge in distribution network state estimation is the unobservable network. Insufficient measurement configuration or interruption of communication channels may lead to an unobservable network.

[0003] An unobservable distribution network refers to a system state that cannot be uniquely determined. Mathematically speaking, when the number of measurements is less than the number of state variables, the state estimation equation is indeterminate and cannot be solved. Existing research on state estimation methods for unobservable distribution networks can be divided into two categories: machine learning and matrix completion. For machine learning methods, a large amount of historical operating data is required, which may be difficult to obtain in practice. In addition, offline training is time-consuming and lacks adaptability to topology changes. Matrix completion refers to supplementing the missing elements in the matrix. Compared with machine learning methods, it only needs to obtain data from a single moment. However, the matrix completion method cannot integrate multi-source measurements into the same matrix well, and due to the use of optimization, the computational burden is large.

[0004] Sudden changes such as distribution line failures caused by natural causes can model the distribution network as a randomly switched subsystem. The state estimation based on the randomly switched subsystem has high estimation accuracy and relatively low computational burden, and has broad application prospects. This subsystem is different from the subsystem in distributed control, where the subsystems run in parallel. The randomly switched subsystem refers to different operating scenarios of the system, and only one subsystem can exist in each time section. State estimation is based on the measurement data from different unobservable subsystems. However, the distribution network usually does not allow frequent switching. Therefore, it is of great significance to achieve state estimation of the unobservable distribution network through a limited number of "active" switching. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and, for the distribution network state estimation problem, establish an unobservable distribution network state estimation method based on a random switching subsystem under unobservable conditions, taking into account the estimation accuracy and calculation speed.

[0006] The present invention solves the technical problem by adopting the following technical solutions:

[0007] The method for estimating the state of an unobservable distribution system based on a random switching subsystem comprises the following steps:

[0008] Step 1: For a given unobservable distribution network, input line impedance parameters, load, access location and capacity of distributed generation, and configuration information of multi-source measurement data, determine the topology structure and state variables of the current period, and set the current period k=0;

[0009] Step 2: Query the operation data of the historical T time sections, and calculate the observability matrix W of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data in step 1. k ; and if the observability matrix W k If the column is full rank, go to step 6; otherwise, go to step 3;

[0010] Step 3: Calculate the measurement dispersion of the current period based on the current period topology in step 1 Constructing a transformable topological set Ω k ;

[0011] Step 4: According to the transformable topology set Ω in step 3 k , calculate the measurement dispersion corresponding to each transformable topology, and select the one with the largest measurement dispersion The corresponding topology is switched, the time period k=k+1 is updated, and the measurement value of the current time period is obtained;

[0012] Step 5: Recalculate the observability matrix of the current period according to the topology switched in step 4, and form a combination matrix according to the observability matrices corresponding to all topologies before time k. If the combination matrix satisfies the full column rank, proceed to step 6; otherwise, return to step 4;

[0013] Step 6: Construct a transfer matrix to transform the state variables in step 1;

[0014] Step 7, using the unscented Kalman filter method to solve the state variables converted in step 6;

[0015] Step 8: Construct an inverse transformation matrix to restore the state variables converted in step 7 and output the state estimation result.

[0016] Moreover, the specific implementation method of the observability matrix in step 2 is:

[0017]

[0018] Where W k is the observability matrix at time k, H k is the measurement state matrix at time k, A k is the state transfer matrix at time k, is the number of state variables.

[0019] Moreover, the calculation method of measuring the dispersion in step 3 is:

[0020]

[0021] In the formula, ρ1, ρ μ , N are the measurement ratios related to nodes 1, μ, and N at time k, μ is the node number, N is the number of nodes, η is the node sequence number, and M k is the measurement matrix at time k, whose main diagonal elements represent the number of node measurements, and the remaining positions represent the number of line measurements, M k (μ,η) represents the matrix M k The μth row and ηth column element of is the incidence matrix at time j, Representation Matrix The μth row and ηth column element of To measure dispersion, var represents the variance operation;

[0022] The transformable topology set in step 3 is:

[0023] Ω k ={1,2,…,M}

[0024] In the formula, Ω k is the transformable topology set at time k, and M is the transformable topology number.

[0025] Furthermore, the maximum dispersion measured in step 4 is The calculation method is:

[0026]

[0027] In the formula, is the maximum measured dispersion, They are the measured dispersions corresponding to the transformable topologies 1, 2, and M respectively.

[0028] Moreover, the combined matrix calculation method in step 5 is:

[0029]

[0030] Where W Θ is the combination matrix, W1, W2, W k are the observability matrices corresponding to the topology at time 1, 2, and k respectively.

[0031] Moreover, the transfer matrix construction method in step 6 is: according to the observability matrix W k Construct the base matrix:

[0032]

[0033] Where Y k is the basis matrix at time k, ker(W k ) is W k The null space of , Base is the basis vector, W k rank; choose any matrix Z k Form the transformation matrix T k , and make T k Reversible:

[0034] T k =[Y k ,Z k ]

[0035]

[0036] Where, T k is the transformation matrix, G k is the up-conversion matrix, F k is the down conversion matrix;

[0037] The state variables after conversion are:

[0038]

[0039] In the formula, x is the original state variable, z is k is the state variable after conversion, v k is the state variable irrelevant term, is the auxiliary state variable;

[0040] The state transfer matrix after conversion and the measurement state matrix It can be expressed as:

[0041]

[0042] In the formula, is the state transfer matrix after conversion, is the measurement state matrix after conversion, The upper state transfer matrix, is the next state transfer matrix, is the right measurement state matrix.

[0043] Moreover, the specific method of step 8 is:

[0044] Φ k =(F' k F k ) -1 F k

[0045]

[0046] In the formula, Φ k is the inverse matrix of the state transition at time k, F k is the conversion matrix at time k, is the estimated value of the state variable after the transformation at time k, x k is the system state variable at time k.

[0047] A device for estimating the state of an unobservable power distribution system based on a random switching subsystem, used for executing a method for estimating the state of an unobservable power distribution system based on a random switching subsystem, comprises a parameter configuration module, an observability matrix construction and judgment module, a measurement dispersion and transformable topology set construction module, a current period measurement value calculation module, a combination matrix construction and judgment module, a state variable conversion module, a state variable calculation module and a state estimation result calculation module, wherein:

[0048] The parameter configuration module is used to input line impedance parameters, load, access location of distributed power sources, capacity and configuration information of multi-source measurement data, determine the state variables of the current time period topology and state estimation, and set the current time period k = 0;

[0049] The observability matrix construction and judgment module is used to query the operation data of the historical T time sections and calculate the observability matrix W of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data. k ; and determine the observability matrix W k Satisfy column full rank;

[0050] The measurement dispersion and transformable topology set construction module is used to calculate the measurement dispersion of the current period according to the topological structure of the current period. Constructing a transformable topological set Ω k ;

[0051] The current period measurement value calculation module is used to calculate the value of the current period measurement value according to the transformable topology set Ω k , calculate the measurement dispersion corresponding to each transformable topology, and select the one with the largest measurement dispersion The corresponding topology is switched, the time period k=k+1 is updated, and the measurement value of the current time period is obtained;

[0052] The combination matrix construction and judgment module is used to recalculate the observability matrix of the current period according to the switched topology, and form a combination matrix according to the observability matrices corresponding to all topologies before time k;

[0053] The state variable conversion module is used to construct the transfer matrix state variables for conversion;

[0054] The state variable calculation module is used to solve the converted state variables using the unscented Kalman filter method;

[0055] The state estimation result calculation module is used to construct the inverse transformation matrix to restore the converted state variables and output the state estimation results.

[0056] A storage medium stores an unobservable distribution system state estimation method based on a random switching subsystem.

[0057] An electronic device implements an unobservable distribution system state estimation method based on random switching subsystems.

[0058] The advantages and positive effects of the present invention are:

[0059] The present invention determines the state variables of state estimation by inputting line impedance parameters, access locations and capacities of loads and distributed power sources, and configuration information of multi-source measurement data; inquiring the operation data of historical T time sections, and calculating the rank of the observability matrix of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data; if the column full rank is not satisfied, the measurement dispersion of the current period is calculated, and the measurement dispersion corresponding to the transformable topology is calculated, and the topology with the largest measurement dispersion is selected for switching to obtain the measurement value of the current period; at the same time, the rank of the combination matrix of the current period is calculated, and if the combination matrix satisfies the column full rank, a transfer matrix is ​​constructed to transform the state variables; the unscented Kalman filter method is used to solve the transformed state variables; the inverse transformation matrix is ​​constructed to restore the transformed state variables, and the state estimation result is output. The present invention models the distribution network as a random switching subsystem under unobservable conditions, reveals the switching path of the distribution network, reduces the frequent switching of the system, maps high-dimensional state variables to low-dimensional conversion variables through state transformation, and uses the improved unscented Kalman filter algorithm for solving, thereby improving the accuracy and speed of state estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flow chart of an unobservable distribution network state estimation method based on a random switching subsystem of the present invention;

[0061] Figure 2 It is an improved IEEE 33-node network topology connection and measurement configuration diagram;

[0062] Figure 3 It is the relationship between MAPE and measurement redundancy;

[0063] Figure 4 It is the comparison diagram of system node voltage amplitude error;

[0064] Figure 5 It is a comparison diagram of the phase angle error of the system node voltage. DETAILED DESCRIPTION

[0065] The present invention is further described in detail below with reference to the accompanying drawings.

[0066] Embodiment 1

[0067] See also Figure 1 , a method for estimating the state of an unobservable distribution system based on a random switching subsystem comprises the following steps:

[0068] Step 1: For a given unobservable distribution network, input line impedance parameters, load, access location and capacity of distributed generation, and configuration information of multi-source measurement data, determine the topology structure and state variables of the current period, and set k=0 for the current period.

[0069] Step 2: Query the operation data of the historical T time sections, and calculate the observability matrix W of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data in step 1. k ; and if the observability matrix W k If the column is full rank, go to step 6; otherwise, go to step 3.

[0070] Query the operation data of the historical T time sections, and calculate the observability matrix W of the current period according to the line impedance parameters in step 1 and the configuration information of the multi-source measurement data k ; The observability matrix calculation method is:

[0071]

[0072] Where W k is the observability matrix at time k, H k is the measurement state matrix at time k, A k is the state transfer matrix at time k, is the number of state variables.

[0073] Step 3: Calculate the measurement dispersion of the current period based on the current period topology in step 1 Constructing a transformable topological set Ω k .

[0074] According to the current period topology in step 1, calculate the measurement dispersion of the current period Constructing a transformable topological set Ω k ;

[0075]

[0076] In the formula, ρ1, ρ μ , N are the measurement ratios related to nodes 1, μ, and N at time k, μ is the node number, N is the number of nodes, η is the node sequence number, and M kis the measurement matrix at time k, whose main diagonal elements represent the number of node measurements, and the remaining positions represent the number of line measurements, M k (μ,η) represents the matrix M k The μth row and ηth column element of is the incidence matrix at time k, Representation Matrix The μth row and ηth column element of To measure dispersion, var represents the variance operation.

[0077] The transformable topology set is:

[0078] Ω k ={1,2,…,M}

[0079] In the formula, Ω k is the transformable topology set at time k, and M is the transformable topology number.

[0080] Step 4: According to the transformable topology set Ω in step 3 k , calculate the measurement dispersion corresponding to each transformable topology, and select the one with the largest measurement dispersion The corresponding topology is switched, the time period k=k+1 is updated, and the measurement value of the current time period is obtained.

[0081] Step 5: According to the topology switched in step 4, recalculate the observability matrix of the current period, and form a combination matrix according to the observability matrices corresponding to all topologies before time k. If the combination matrix satisfies the full column rank, proceed to step 6; otherwise, return to step 4.

[0082] According to the switched topology, the observability matrix of the current period is recalculated, and a combination matrix is ​​formed according to the observability matrices corresponding to all topologies before time k. The calculation method of the combination matrix is:

[0083]

[0084] Where W Θ is the combination matrix, W1, W2, W k are the observability matrices corresponding to the topology at time 1, 2, and k respectively.

[0085] Step 6: Construct a transfer matrix to transform the state variables in step 1.

[0086] According to the observability matrix W k Construct the base matrix:

[0087]

[0088] Where Y k is the basis matrix at time k, ker(Wk ) is W k The null space of , Base is the basis vector, W k rank.

[0089] Choose any matrix Z k Form the transformation matrix T k , and make T k Reversible:

[0090] T k =[Y k ,Z k ] (6)

[0091]

[0092] Where, T k is the transformation matrix, G k is the up-conversion matrix, F k is the down conversion matrix.

[0093] The state variables after conversion are:

[0094]

[0095] In the formula, x is the original state variable, z is k is the state variable after conversion, v k is the state variable irrelevant term, Auxiliary state variables.

[0096] The state transfer matrix after conversion and the measurement state matrix It can be expressed as:

[0097]

[0098] In the formula, is the state transfer matrix after conversion, is the measurement state matrix after conversion, The upper state transfer matrix, is the next state transfer matrix, is the right measurement state matrix.

[0099] Step 7: Use the unscented Kalman filter method to solve the state variables converted in step 6.

[0100] The unscented Kalman filter method is used to solve the transformed state variables;

[0101] First select The group has the same sigma point as the state variable with the same mean and variance. The sigma point and its weight are calculated as:

[0102]

[0103] In the formula, is the sigma point set, is the average value of the state variable in period k, is the distance control parameter, P k is the state covariance matrix of time period k, is the mean weight, is the covariance weight.

[0104] State propagation is performed for each group of sigma points, and the mean and covariance matrix of the propagated sigma points are calculated as follows:

[0105]

[0106] In the formula, is the prior value of the transition state variable for period k+1, is the mean of the prior values ​​of the transition state variables in period k+1, is the prior value of the state covariance matrix of time period k, Q k is the forecast covariance matrix for period k.

[0107] Based on the propagated sigma point, perform unscented transformation again to obtain a new sigma point. The new sigma point is state propagated again, and the mean and covariance matrix are calculated, which can be expressed as:

[0108]

[0109] In the formula, is the new sigma point set, is the measured predicted value of period k+1, is the measurement covariance matrix for period k, is the measurement state covariance matrix of time period k, R k is the measurement error matrix.

[0110] Calculate the Kalman gain coefficient and update the converted state variables and covariance matrix, which can be expressed as:

[0111]

[0112] In the formula, K k+1 is the Kalman gain coefficient for time period k+1, is the posterior value of the state covariance matrix for period k+1.

[0113] Step 8: Construct an inverse transformation matrix to restore the state variables converted in step 7 and output the state estimation result.

[0114] Construct the inverse transformation matrix to restore the transformed state variables and output the state estimation results. The specific expression is:

[0115] Φ k =(F' k F k ) -1 F k (twenty four)

[0116]

[0117] In the formula, Φ k is the inverse matrix of the state transition at time k, F k is the conversion matrix at time k, is the estimated value of the state variable after the transformation at time k, x k is the system state variable at time k.

[0118] According to the above-mentioned unobservable distribution system state estimation method based on random switching subsystem, the improved IEEE33-node distribution network example is used to verify the method proposed in the present invention. The network topology and measurement configuration are as follows: Figure 2 As shown in the figure: nodes 7, 16, 41, and 47 are connected to a four-terminal flexible distribution device with a capacity of 3.0MA per port; nodes 5, 9, 16, 18, 20, and 24 are connected to 6 photovoltaic groups; nodes 10, 11, 21, 29, 17, 31, and 33 are connected to 7 wind turbines. The unobservable distribution network state estimation method based on random switching subsystem is used for state estimation calculation. After the above steps, the operation state of the distribution network can be obtained. To verify the effectiveness of this method, the following three schemes are used for comparison.

[0119] Solution 1: Using the traditional Gauss-Newton method (the method proposed by the present invention);

[0120] Solution 2: adopting the unobservable distribution network state estimation method based on random switching subsystem of the present invention;

[0121] Solution 3: Use matrix completion method.

[0122] The computer hardware environment for performing optimization calculations is Core i7, with a main frequency of 3.20GHz and a memory of 16GB; the software environment is the Windows 11 operating system.

[0123] Using the unobservable distribution network state estimation method based on random switching subsystems described in the present invention, the relationship between measurement redundancy and MAPE is as follows: Figure 3As shown in the figure, it can be seen that when the measurement redundancy is greater than 1 (the system is observable), the estimation accuracy of Scheme 1 and Scheme 2 is at the same level, but both are better than Scheme 4. When the measurement redundancy is less than 1 (the system is unobservable), Scheme 1 is no longer applicable, and Scheme 2 is better than Scheme 3, which proves that the proposed method can still guarantee high-precision state estimation even under unobservable conditions, and can effectively improve the perception ability of the distribution network.

[0124] Table 1 Comparison of state estimation errors under unobservable conditions

[0125]

[0126] Table 1 and Figure 4-Figure 5 The state estimation results under unobservable conditions are given. In particular, Scheme 2 is at the same level of accuracy as Scheme 3 in terms of voltage amplitude, but is significantly better than Scheme 3 in terms of voltage phase angle.

[0127] Table 2 Comparison of calculation time

[0128]

[0129] Table 2 lists the state estimation calculation time of different schemes under different conditions. Due to solving the optimization problem, the matrix completion method used in Scheme 3 takes the longest time. The calculation speed of Scheme 2 is nearly twice that of Scheme 4, which effectively ensures the real-time performance of state estimation.

[0130] Embodiment 2

[0131] A device for estimating the state of an unobservable power distribution system based on a random switching subsystem, used to execute the above-mentioned method for estimating the state of an unobservable power distribution system based on a random switching subsystem, comprises a parameter configuration module, an observability matrix construction and judgment module, a measurement dispersion and transformable topology set construction module, a current period measurement value calculation module, a combination matrix construction and judgment module, a state variable conversion module, a state variable calculation module and a state estimation result calculation module, wherein:

[0132] The parameter configuration module is used to input line impedance parameters, load, access location of distributed power sources, capacity and configuration information of multi-source measurement data, determine the state variables of the current time period topology and state estimation, and set the current time period k = 0;

[0133] The observability matrix construction and judgment module is used to query the operation data of the historical T time sections and calculate the observability matrix W of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data. k ; and determine the observability matrix W k Satisfy column full rank;

[0134] The measurement dispersion and transformable topology set construction module is used to calculate the measurement dispersion of the current period according to the topological structure of the current period. Constructing a transformable topological set Ω k ;

[0135] The current period measurement value calculation module is used to calculate the value of the current period measurement value according to the transformable topology set Ω k , calculate the measurement dispersion corresponding to each transformable topology, and select the one with the largest measurement dispersion The corresponding topology is switched, the time period k=k+1 is updated, and the measurement value of the current time period is obtained;

[0136] The combination matrix construction and judgment module is used to recalculate the observability matrix of the current period according to the switched topology, and form a combination matrix according to the observability matrices corresponding to all topologies before time k;

[0137] The state variable conversion module is used to construct the transfer matrix state variables for conversion;

[0138] The state variable calculation module is used to solve the converted state variables using the unscented Kalman filter method;

[0139] The state estimation result calculation module is used to construct the inverse transformation matrix to restore the converted state variables and output the state estimation results.

[0140] Embodiment 3

[0141] A storage medium is used to store an unobservable distribution system state estimation method based on a random switching subsystem.

[0142] Embodiment 4

[0143] An electronic device is used to implement an unobservable distribution system state estimation method based on randomly switched subsystems.

[0144] It should be emphasized that the embodiments described in the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation manner. Any other implementation manners derived by those skilled in the art based on the technical solution of the present invention also fall within the scope of protection of the present invention.

Claims

1. An unobservable distribution system state estimation method based on random switching subsystems, characterized by: The following steps are involved: Step 1: For a given unobservable distribution network, input line impedance parameters, load, access location and capacity of distributed generation, and configuration information of multi-source measurement data, determine the topology structure and state variables of the current period, and set the current period k=0; Step 2: Query the operation data of the historical T time sections, and calculate the observability matrix W of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data in step 1. k ; and if the observability matrix W k If the column is full rank, go to step 6; otherwise, go to step 3; Step 3: Calculate the measurement dispersion of the current period based on the current period topology in step 1 Constructing a transformable topological set Ω k ; Step 4: According to the transformable topology set Ω in step 3 k , calculate the measurement dispersion corresponding to each transformable topology, and select the one with the largest measurement dispersion The corresponding topology is switched, the time period k=k+1 is updated, and the measurement value of the current time period is obtained; Step 5: Recalculate the observability matrix of the current period according to the topology switched in step 4, and form a combination matrix according to the observability matrices corresponding to all topologies before time k. If the combination matrix satisfies the full column rank, proceed to step 6; otherwise, return to step 4; Step 6: Construct a transfer matrix to transform the state variables in step 1; Step 7, using the unscented Kalman filter method to solve the state variables converted in step 6; Step 8: Construct an inverse transformation matrix to restore the state variables converted in step 7 and output the state estimation result.

2. The unobservable distribution system state estimation method based on random switching subsystem according to claim 1 is characterized in that: The specific implementation method of the observability matrix in step 2 is: Where W k is the observability matrix at time k, H k is the measurement state matrix at time k, A k is the state transfer matrix at time k, is the number of state variables.

3. The unobservable distribution system state estimation method based on random switching subsystem according to claim 1 is characterized in that: The method for calculating the measured dispersion in step 3 is: In the formula, ρ1, ρ μ , N are the measurement ratios related to nodes 1, μ, and N at time k, μ is the node number, N is the number of nodes, η is the node sequence number, and M k is the measurement matrix at time k, whose main diagonal elements represent the number of node measurements, and the remaining positions represent the number of line measurements, M k (μ,η) represents the matrix M k The μth row and ηth column element of is the incidence matrix at time k, Representation Matrix The μth row and ηth column element of To measure dispersion, var represents the variance operation; The transformable topology set in step 3 is: Oh k ={1,2,…,M} In the formula, Ω k is the transformable topology set at time k, and M is the transformable topology number.

4. The unobservable distribution system state estimation method based on random switching subsystem according to claim 1 is characterized in that: The maximum dispersion measured in step 4 The calculation method is: In the formula, is the maximum measured dispersion, They are the measured dispersions corresponding to the transformable topologies 1, 2, and M respectively.

5. The unobservable distribution system state estimation method based on random switching subsystem according to claim 1 is characterized in that: The combined matrix calculation method in step 5 is: Where W Θ is the combination matrix, W1, W2, W k are the observability matrices corresponding to the topology at time 1, 2, and k respectively.

6. The unobservable distribution system state estimation method based on random switching subsystem according to claim 1 is characterized in that: The transfer matrix construction method in step 6 is: according to the observability matrix W k Construct the base matrix: Where Y k is the basis matrix at time k, ker(W k ) is W k The null space of , Base is the basis vector, W k rank; Choose any matrix Z k Form the transformation matrix T k , and make T k Reversible: T k =[Y k ,Z k ] Where, T k is the transformation matrix, G k is the up-conversion matrix, F k is the down conversion matrix; The state variables after conversion are: In the formula, x is the original state variable, z is k is the state variable after conversion, v k is the state variable irrelevant term, is the auxiliary state variable; The converted state transfer matrix and the measurement state matrix It can be expressed as: In the formula, is the state transfer matrix after conversion, is the measurement state matrix after conversion, The upper state transfer matrix, is the next state transfer matrix, is the right measurement state matrix.

7. The unobservable distribution system state estimation method based on random switching subsystem according to claim 1 is characterized by: The specific method of step 8 is: Φ k =(F' k F k ) -1 F k In the formula, Φ k is the inverse matrix of the state transition at time k, F k is the conversion matrix at time k, is the estimated value of the state variable after the transformation at time k, x k is the system state variable at time k.

8. A device for estimating the state of an unobservable power distribution system based on a random switching subsystem, used to execute the method for estimating the state of an unobservable power distribution system based on a random switching subsystem as described in any one of claims 1 to 7, comprising a parameter configuration module, an observability matrix construction and judgment module, a measurement dispersion and transformable topology set construction module, a current period measurement value calculation module, a combination matrix construction and judgment module, a state variable conversion module, a state variable calculation module and a state estimation result calculation module, wherein: The parameter configuration module is used to input line impedance parameters, load, access location of distributed power sources, capacity and configuration information of multi-source measurement data, determine the state variables of the current time period topology and state estimation, and set the current time period k = 0; The observability matrix construction and judgment module is used to query the operation data of the historical T time sections and calculate the observability matrix W of the current period according to the line impedance parameters and the configuration information of the multi-source measurement data. k ; and determine the observability matrix W k Satisfy column full rank; The measurement dispersion and transformable topology set construction module is used to calculate the measurement dispersion of the current period according to the topological structure of the current period. Constructing a transformable topological set Ω k ; The current period measurement value calculation module is used to calculate the value of the current period measurement value according to the transformable topology set Ω k , calculate the measurement dispersion corresponding to each transformable topology, and select the one with the largest measurement dispersion The corresponding topology is switched, the time period k=k+1 is updated, and the measurement value of the current time period is obtained; The combination matrix construction and judgment module is used to recalculate the observability matrix of the current period according to the switched topology, and form a combination matrix according to the observability matrices corresponding to all topologies before time k; The state variable conversion module is used to construct the transfer matrix state variables for conversion; The state variable calculation module is used to solve the converted state variables using the unscented Kalman filter method; The state estimation result calculation module is used to construct the inverse transformation matrix to restore the converted state variables and output the state estimation results.

9. A storage medium storing the unobservable distribution system state estimation method based on random switching subsystems as described in any one of claims 1 to 7.

10. An electronic device, executing the unobservable distribution system state estimation method based on random switching subsystem as described in any one of claims 1 to 7.

Images