Digital-analog double-drive three-phase unbalanced power distribution network interval state estimation method, program, equipment and storage medium

Abstract

The invention belongs to the technical field of power distribution network state estimation, and particularly relates to a digital-analog double-drive three-phase unbalanced power distribution network interval state estimation method, program and device and a storage medium. According to the method, a power distribution network interval state estimation model under the three-phase imbalance is established, a mathematical solution method of affine operation is introduced, and interval state estimation with higher precision under the three-phase imbalance can be realized. According to the method, the fitting and rapid solving characteristics of the neural network are utilized, and rapid and accurate state estimation is realized based on a mathematical physical model. Compared with an interval state estimation method of a pure mathematical physical model, the machine learning method based on data driving provided by the invention can greatly reduce the calculation time of state estimation and improve the efficiency of state estimation while ensuring the accuracy of an estimated value.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network state estimation, and specifically relates to a method, program, device and storage medium for estimating the interval state of a three-phase unbalanced distribution network with digital-analog dual drive. Background Art

[0002] As distributed power sources are increasingly connected to the power distribution system, the operation of the power distribution system has become more complex. In order to accurately and effectively monitor and control the power distribution system, it is necessary to estimate the state of the power distribution system. How to accurately obtain the real-time state of the system operation from uncertain measurement data has become a problem that needs to be solved urgently. Summary of the invention

[0003] The purpose of the present invention is to provide a method, program, device and storage medium for estimating the interval state of a three-phase unbalanced distribution network with digital and analog dual drive, which overcomes the problems of traditional interval state estimation methods being sensitive to data and taking a long time, and can achieve accurate and rapid solution of state estimation.

[0004] A method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive comprises the following steps:

[0005] Step 1: Obtain historical measurement data of the distribution network, including the voltage and injected power of each node in the distribution network, and the current and power of each branch; fill in the bad data or missing data in the historical measurement data through the interval prediction model to obtain pseudo measurement data in interval form; the interval form is β is the lower bound of β, is the upper bound of β;

[0006] Step 2: Perform measurement transformation on the voltage of each node and the current measurement data of each branch in the historical measurement data of the distribution network and the pseudo measurement data in interval form, and fuse the transformed data to obtain the voltage of each node and each phase in the distribution network in interval form. The real part of the voltage and the imaginary part of voltage Each branch ij each phase The real part of the current and the imaginary part of current

[0007] The historical measurement data of the distribution network and the injected power of each node and the power of each branch in the pseudo measurement data in interval form are equivalently transformed, and the equivalent transformed data are integrated to obtain the power of each phase of each node i in the interval distribution network. The real part of the injected current The imaginary part of the injected current Each branch ij each phase The real part of the current and the imaginary part of current Will As an interval measurement vector;

[0008] Step 3: According to the mathematical model of three-phase unbalanced interval state estimation of the distribution network, solve the interval state estimation vector corresponding to each group of interval measurement vectors [z] is the node i in the interval distribution network The estimated value of the real part of the voltage of the phase, is the node i in the interval distribution network Estimated value of the imaginary part of the phase voltage;

[0009] Step 4: Take each group ([z], [x]) as a training sample and train the neural network so that the neural network can learn the relationship between [z] and [x];

[0010] Step 5: Obtain the measurement data of the distribution network for which interval state estimation is to be performed, obtain the measurement vector [z] after processing in step 2, input the measurement vector [z] into the trained neural network, and obtain the interval state estimation vector [x] of the distribution network.

[0011] Furthermore, in step 1, the bad data or missing data in the historical measurement data is filled by the interval prediction model to obtain pseudo measurement data in the form of intervals, specifically:

[0012] Use XGBoost to calculate the measured data [β1,β2,...,β 24 ] Learn to predict and fill in the bad or missing data at the 25th timestamp β 25 , assuming that β 25 Obeying Gaussian distribution, given the confidence level 1-α, the pseudo-measurement data in interval form [β 25 ]for:

[0013]

[0014] Among them, μ and σ are β 25 The expectation and standard deviation of the posterior probability; Z α / 2 is the parameter corresponding to the confidence level 1-α.

[0015] Furthermore, in step 1, the voltage of each node in the distribution network and the current of each branch are measured by PMU; the injected power of each node in the distribution network and the power of each branch are measured by SCADA;

[0016] In step 2, the voltage of each node and the current of each branch measured by the PMU are measured and transformed, specifically:

[0017]

[0018] in, is the PMU measurement at node i Phase voltage amplitude; is the PMU measurement at node i The voltage phase angle of the phase; The branch ij measured by the PMU Phase current amplitude; The branch ij measured by the PMU The current phase angle of the phase;

[0019] The injected power of each node and the power of each branch measured by SCADA are equivalently converted as follows:

[0020]

[0021] in, is the node i of SCADA measurement The injected active power of the phase; is the node i of SCADA measurement The injected reactive power of the phase; The branch ij measured by SCADA Active power of each phase; For SCADA

[0022] The measured branch ij Reactive power of each phase;

[0023] Except for the pseudo-measurement data in interval form, the rest of the measurement data β is converted to interval form season

[0024] Furthermore, the step 3 is specifically as follows:

[0025] Mathematical model for three-phase unbalanced interval state estimation of distribution network:

[0026] [z]=[H][x]+[ε]

[0027] Where [H] is the Jacobian matrix in interval form; [ε] is the error term in interval form;

[0028] The weighted least squares method is used to introduce the weight matrix W and the transition phase [y] = [H] [x] - [z], and the mathematical model of the three-phase unbalanced interval state estimation of the distribution network is converted into:

[0029] [A][X]=[B]

[0030]

[0031] Among them, I is the unit matrix;

[0032] The Krawczyk algorithm is used to iteratively obtain the nonlinear interval state variables:

[0033] [X k+1 ]=[K(x)]∩[X k ]

[0034] [K(x)]=C[B]+(IC[A])

[0035]

[0036] Wherein, Mid[A] means taking the middle value of the interval elements in the matrix [A]; k is the number of iterations; n is the number of nodes in the distribution network; m is the measurable quantity;

[0037] when and Or when the number of iterations k reaches the preset upper limit of the number of iterations, stop the iteration and let Then the interval state estimation vector [x] is obtained.

[0038] Furthermore, when calculating C[B]+(IC[A]), an affine operation is introduced to reduce the interval expansion, and the interval matrices [B] and [A] are converted into affine matrices calculate After the calculation is completed, Convert to interval number [K(x)], and then solve [X k+1 ];

[0039] The conversion relationship between interval numbers and affine numbers is:

[0040] Define the number of intervals Affine number ε i =[-1,1] indicates the noise source;

[0041] Convert interval numbers to affine numbers:

[0042] Convert affine numbers to interval numbers:

[0043] Furthermore, the method for solving the interval-form Jacobian matrix [H] is:

[0044]

[0045] Among them, R, S, and T are the node voltage measurement set, node current measurement set, and branch current measurement set in the distribution network respectively; i,j=1,2,...,n;

[0046] Generate the three-phase admittance matrix Y of the distribution network B :

[0047]

[0048] Among them, Y ij represents the admittance matrix of branch ij; is the mutual admittance of the three-phase branch ij; is the self-admittance of the three-phase branch ij; is the mutual conductance between the three phases of branch ij; It is the self-conduction between the three phases of branch ij; is the mutual acceptance between the three phases of branch ij; is the self-susceptance between the three phases of branch ij;

[0049] The functional relationship between the amount of node injection current measurement and the state quantity is:

[0050]

[0051] Where Φ = {a, b, c}; Ω is the set of nodes connected to node i; and They are respectively Conductance and susceptance relative to the φ phase;

[0052] The functional relationship between the branch current measurement and the state quantity is:

[0053]

[0054] Based on the above two sets of functional relationships, the elements in the Jacobian matrix [H] are solved.

[0055] Furthermore, in step 4, the neural network is trained using the loss function:

[0056] Loss = α Loss1 + (1-α) Loss2

[0057] Among them, α is a hyperparameter that adjusts the two loss functions;

[0058]

[0059] Among them, x is the input of the neural network, y is the output of the neural network; δ is the parameter of the loss function;

[0060] For the loss function in the training process, physical constraints are introduced to prevent the training results from deviating from the normal range; the per-unit values ​​of the real and imaginary parts of the voltage at each node in the distribution network are in the interval [V re,low ,V re,up ]、[V im,low ,V im,up] remain stable, and this interval is determined according to the specific distribution network system;

[0061]

[0062] in, are the upper and lower bounds of the interval estimator of the real part of the voltage at node i, respectively; are the upper and lower bounds of the interval estimator of the imaginary part of the voltage at node i, respectively.

[0063] A computer device / equipment / system comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-mentioned method for estimating the interval state of a three-phase unbalanced distribution network with digital-analog dual drive.

[0064] A computer-readable storage medium stores a computer program / instruction thereon, which, when executed by a processor, implements the steps of the above-mentioned method for estimating the interval state of a three-phase unbalanced distribution network with digital-analog dual drive.

[0065] A computer program product includes a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned method for estimating the interval state of a three-phase unbalanced distribution network with digital-analog dual drive.

[0066] The beneficial effects of the present invention are:

[0067] The present invention establishes a distribution network interval state estimation model under three-phase imbalance, introduces a mathematical solution method of affine operation, and can achieve higher-precision interval state estimation under three-phase imbalance. The present invention utilizes the characteristics of neural network fitting and fast solution, relying on mathematical and physical models, to achieve fast and accurate state estimation. Compared with the interval state estimation method of pure mathematical and physical models, the data-driven machine learning method provided by the present invention can greatly reduce the calculation time of state estimation while ensuring the accuracy of the estimated value, thereby improving the efficiency of state estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 The present invention provides a flow chart of a method for estimating the interval state of a three-phase unbalanced distribution network with digital-analog dual drive.

[0069] Figure 2 It is a schematic diagram of the measurement of neural network learning amount and interval state quantity in the present invention.

[0070] Figure 3 It is a schematic diagram of realizing three-phase interval state estimation by the neural network in the present invention.

[0071] Figure 4 Schematic diagram of the loss function composition of the neural network training in the present invention. DETAILED DESCRIPTION

[0072] The present invention is further described below in conjunction with the accompanying drawings.

[0073] The present invention provides a method for estimating interval state of a three-phase unbalanced distribution network with a digital-analog dual-drive that combines physical models and data drive. The uncertainty of the measurement data is taken into account. First, XGBoost is used to perform interval prediction for bad measurement data or missing data to obtain pseudo measurement data in interval form. Then, the data measured by SCADA and PMU are fused to obtain equivalent measurement information. Then, an interval solution algorithm is used to solve the interval-form three-phase unbalanced state estimation model to obtain interval state quantities. Finally, a neural network is used to learn the relationship between measurement quantities and state quantities to achieve rapid solution of state estimation. Figure 1 As shown, the present invention comprises the following steps:

[0074] Step 1: Obtain historical measurement data of the distribution network, including the voltage of each node and the current of each branch in the distribution network measured by PMU, and the injected power of each node in the distribution network and the power of each branch measured by SCADA;

[0075] For bad or missing data in historical measurement data, XGBoost is used to extract the measurement data [β1,β2,...,β 24 ] Learn to predict and fill in the bad or missing data at the 25th timestamp β 25 , and obtain pseudo-measurement data in interval form β is the lower bound of β, is the upper bound of β;

[0076] Assuming that it follows a Gaussian distribution, given a confidence level of 1-α, the confidence interval of the prediction result is:

[0077]

[0078] Among them, μ and σ are β 25 The expectation and standard deviation of the posterior probability; Z α / 2 is the parameter corresponding to the confidence level 1-α. When the confidence level is 95%, Z α / 2 =1.96.

[0079] Step 2: Transform the voltage of each node and the current of each branch measured by the PMU to obtain the phase of each node i in the distribution network in interval form. The real part of the voltage and the imaginary part of voltage Each branch ij each phase The real part of the current and the imaginary part of current

[0080]

[0081] in, is the PMU measurement at node i Phase voltage amplitude; is the PMU measurement at node i The voltage phase angle of the phase; The branch ij measured by the PMU Phase current amplitude; The branch ij measured by the PMU The current phase angle of the phase;

[0082] The injected power of each node and the power of each branch measured by SCADA are equivalently converted to obtain the power of each phase of each node i in the interval distribution network. The real part of the injected current The imaginary part of the injected current Each branch ij each phase The real part of the current and the imaginary part of current

[0083]

[0084] in, is the node i of SCADA measurement The injected active power of the phase; is the node i of SCADA measurement The injected reactive power of the phase; The branch ij measured by SCADA Active power of each phase; The branch ij measured by SCADA Reactive power of each phase; All with and The functional form of

[0085] The pseudo-measurement data in interval form is also transformed by equations (1), (2), (3), and (4) to obtain the real and imaginary parts of the voltage and current in interval form:

[0086] Except for the pseudo-measurement data in interval form, the rest of the measurement data β is converted to interval form season

[0087] All data converted from the above PMU measurements, SCADA measurements and pseudo-measurement data in interval form are integrated; As the interval state estimation vector, is the node i in the interval distribution network The estimated value of the real part of the voltage of the phase, is the node i in the interval distribution network The estimated value of the imaginary part of the voltage of the phase; As an interval measurement vector;

[0088] Step 3: According to the mathematical model of three-phase unbalanced interval state estimation of the distribution network, solve the interval state estimation vector corresponding to each group of interval measurement vectors [z] is the node i in the interval distribution network The estimated value of the real part of the voltage of the phase, is the node i in the interval distribution network Estimated value of the imaginary part of the phase voltage;

[0089] The general expression of the three-phase unbalanced interval measurement equation is:

[0090] [z]=[H][x]+[ε] (5)

[0091] Where [H] is the Jacobian matrix in interval form; [ε] is the error term in interval form;

[0092] According to the quantity measurement and state quantity in the present invention, formula (5) can be expanded as follows:

[0093]

[0094] Among them, R, S, and T are the node voltage measurement set, node current measurement set, and branch current measurement set in the distribution network respectively; i,j=1,2,...,n;

[0095] Generate the three-phase admittance matrix Y of the distribution network B :

[0096]

[0097] Among them, Y ij represents the admittance matrix of branch ij; is the mutual admittance of the three-phase branch ij; is the self-admittance of the three-phase branch ij; is the mutual conductance between the three phases of branch ij; It is the self-conduction between the three phases of branch ij; is the mutual acceptance between the three phases of branch ij; is the self-susceptance between the three phases of branch ij;

[0098] The functional relationship between the amount of node injection current measurement and the state quantity is:

[0099]

[0100] Where Φ = {a, b, c}; Ω is the set of nodes connected to node i; and They are respectively Conductance and susceptance relative to the φ phase;

[0101] The functional relationship between the branch current measurement and the state quantity is:

[0102]

[0103] From formulas (8) and (9), the elements of the Jacobian matrix [H] can be found.

[0104] According to the weighted least squares method, formula (5) can be converted into:

[0105]

[0106] Where W is the weight matrix, which is the inverse of the covariance matrix of the measurement error, σ i is the standard deviation of the i-th quantity measurement.

[0107] To avoid [H] T A large number of interval calculations in W[H] introduce the transition phase [y] = [H][x] - [z], and equation (10) is converted into

[0108]

[0109] Formula (11) is equivalent to:

[0110] [A][X]=[B] (12)

[0111]

[0112] In step 4, for equation (12), the Krawczyk algorithm is used to iteratively obtain the nonlinear interval state variable:

[0113]

[0114] Where:

[0115]

[0116] Among them, Mid[A] means taking the middle value of the interval elements in the matrix [A]; k is the number of iterations; n is the number of nodes in the distribution network; and m is the measurable quantity.

[0117] When calculating C[B]+(IC[A]), an affine operation is introduced to reduce interval expansion and the interval matrices [B] and [A] are converted into affine matrices. calculate After the calculation is completed, Convert it into interval number [K(x)], substitute it into formula (14) and solve [X k+1 ];

[0118] The conversion relationship between interval numbers and affine numbers is:

[0119] Define the number of intervals Affine number ε i =[-1,1] indicates the noise source;

[0120] Convert interval numbers to affine numbers:

[0121] Convert affine numbers to interval numbers:

[0122] Step 4: Take each group ([z], [x]) as a training sample, train the neural network, and make the neural network learn the relationship between [z] and [x]. Figure 2 A simple fully connected neural network is shown as an example, and the neural network is used to learn the relationship between the measured value and the estimated interval value.

[0123] The Huber Loss loss function used in the present invention combines the advantages of the mean squared error (MSE) loss function and the mean absolute error (MAE) loss function, reduces the sensitivity to outliers, and realizes the function of being differentiable everywhere.

[0124]

[0125] Where: x is the value output by the neural network, and y is the interval state quantity obtained by solving the physical model.

[0126] When |xy|≤δ, it becomes MSE; when When , it becomes similar to MAE. The advantage of Huber Loss is that it is robust to outliers and avoids the disadvantages of MSE and MAE. In practical applications, the appropriate δ value can be selected according to the specific situation of the problem.

[0127] For the loss function in the training process, physical constraints are introduced to prevent the training results from deviating from the normal range and violating the laws of physics. Usually, the per-unit values ​​of the real and imaginary parts of the voltage at each node in the power system are in the range [V re,low ,V re,up]、[V im,low ,V im,up ] remains stable, and this interval can be determined according to the specific system.

[0128]

[0129] Where: are the upper and lower bounds of the interval estimator of the real part of the voltage at node i, respectively; are the upper and lower bounds of the interval estimator of the imaginary part of the voltage at node i, respectively;

[0130] Finally, the loss function during network training is:

[0131]

[0132] Among them, α is a hyperparameter that adjusts the two loss functions;

[0133] Step 5: Obtain the measurement data of the distribution network for which interval state estimation is to be performed, obtain the measurement vector [z] after processing in step 2, input the measurement vector [z] into the trained neural network, and obtain the interval state estimation vector [x] of the distribution network.

[0134] Embodiment 1:

[0135] IEEE33 node implementation example:

[0136] Step 1: Obtain node voltage, node injection power, branch power, and branch current from the SCADA system, and obtain node voltage and branch current information measured by the PMU.

[0137]

[0138] The acquired data is cleaned to obtain the original data with missing data. XGBoost is used to measure [z1,z2,…,z 23 ,z 24 ]Fill in missing or bad data at the 25th timestamp 25 , and considering the 95% confidence level, it is transformed into an interval pseudo-measure [z 25 ].

[0139] Step 2: Perform data fusion on historical measurement data and pseudo measurement data measured by different types of equipment;

[0140] Step 3: Establish and solve the three-phase unbalanced interval state estimation model after data mixing;

[0141] Step 3.1, establish the measurement equation;

[0142] Step 3.2, find the Jacobian matrix [H];

[0143] Elements in the Jacobian matrix related to the real and imaginary parts of the equivalent node voltage:

[0144]

[0145] The other elements are all 0.

[0146] Elements in the Jacobian matrix related to the equivalent node injection current:

[0147]

[0148] When k∈Ω,

[0149]

[0150] The other elements are all 0.

[0151] Current in equivalent branch:

[0152]

[0153] The other elements are all 0.

[0154] According to equations (6)(19)(20)(21)(22), find the elements in the Jacobian matrix [H].

[0155] Step 3.3, using the interval solution algorithm to solve the state estimation model;

[0156] Establish the least squares solution model of equations (10)(11)(12)(13).

[0157] Step 4, using Krawczyk algorithm and affine number iteration to obtain state variables;

[0158] Step 4.1 Write the Krawczyk algorithm iteration formula according to equation (14). In the initial iteration interval, the real part of the state variable is [0.90, 1.10], the imaginary part is [-0.01, 0.02], and the transition phase [y] is [0, 0].

[0159] Step 4.2: When calculating [K(x)] in equation (14), use affine operation instead of interval operation to reduce the over-conservatism caused by interval expansion. According to the conversion relationship between interval number and affine number, C[B]+(IC[A]) is converted to Perform an affine operation and then convert the result of the affine operation into an interval number.

[0160] Step 4.3: When the convergence condition is met, stop the iteration and obtain the interval state quantity.

[0161] Step 5: Create a neural network model and train the neural network to achieve fast state estimation.

[0162] Step 5.1, select a three-layer fully connected neural network, and use the neural network to learn the relationship between the measured value and the estimated interval value, such as Figure 2 The input data does not use pseudo-quantity measurement, only directly measured data is used, and the output data is in the form of the upper and lower bounds of the real part of the interval state quantity and the upper and lower bounds of the imaginary part.

[0163] Step 5.2: Establish the loss function;

[0164] Step 5.2.1. Select the HuberLoss loss function as shown in formula (16), and the value of δ is 0.5.

[0165] Step 5.2.2: Create a physical loss function as shown in equation (17). Physical constraints are introduced to prevent the training results from deviating from the normal range and violating the laws of physics. The per-unit values ​​of the real and imaginary parts of the voltage at each node in the power system are in the interval [V re,low ,V re,up ]=[0.90,1.10][V im,low ,V im,up ]=[0,0.02] and remain stable. This interval can be determined according to the specific system.

[0166] Step 5.2.3: Generate the loss function for the final neural network training, such as Figure 4 , as shown in formula (18): where α is taken as 0.8. Step 5.3. Set other hyper parameters for neural network training, as shown in the following table.

[0167]

[0168] Step 5.4: Train the neural network and provide the processed real-time measurements to the trained network model to achieve accurate and fast interval state estimation.

[0169] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for estimating the interval state of a three-phase unbalanced distribution network with digital-analog dual drive, characterized in that: The following steps are involved: Step 1: Obtain historical measurement data of the distribution network, including the voltage and injected power of each node in the distribution network, and the current and power of each branch; fill in the bad data or missing data in the historical measurement data through the interval prediction model to obtain pseudo measurement data in interval form; the interval form is β is the lower bound of β, is the upper bound of β; Step 2: Perform measurement transformation on the voltage of each node and the current measurement data of each branch in the historical measurement data of the distribution network and the pseudo measurement data in interval form, and fuse the transformed data to obtain the voltage of each node and each phase in the distribution network in interval form. The real part of the voltage and the imaginary part of voltage Each branch ij each phase The real part of the current and the imaginary part of current The historical measurement data of the distribution network and the injected power of each node and the power of each branch in the pseudo measurement data in interval form are equivalently transformed, and the equivalent transformed data are integrated to obtain the power of each phase of each node i in the distribution network in interval form. The real part of the injected current The imaginary part of the injected current Each branch ij each phase The real part of the current and the imaginary part of current Will As an interval measurement vector; Step 3: According to the mathematical model of three-phase unbalanced interval state estimation of the distribution network, solve the interval state estimation vector corresponding to each group of interval measurement vectors [z] is the node i in the distribution network in interval form The estimated value of the real part of the voltage of the phase, is the node i in the interval distribution network Estimated value of the imaginary part of the voltage of the phase; Step 4: Take each group ([z], [x]) as a training sample and train the neural network so that the neural network can learn the relationship between [z] and [x]; Step 5: Obtain the measurement data of the distribution network for which interval state estimation is to be performed, obtain the measurement vector [z] after processing in step 2, input the measurement vector [z] into the trained neural network, and obtain the interval state estimation vector [x] of the distribution network.

2. The method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive according to claim 1 is characterized by: In step 1, the bad data or missing data in the historical measurement data are filled by the interval prediction model to obtain pseudo measurement data in the form of intervals, specifically: Use XGBoost to calculate the measured data [β1,β2,...,β 24 ] Learn to predict and fill in the bad or missing data at the 25th timestamp β 25 , assuming β 25 Obeying Gaussian distribution, given the confidence level 1-α, the pseudo-measurement data in interval form [β 25 ]for: Among them, μ and σ are β 25 The expectation and standard deviation of the posterior probability; Z α / 2 is the parameter corresponding to the confidence level 1-α.

3. The method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive according to claim 1 is characterized in that: In step 1, the voltage of each node in the distribution network and the current of each branch are measured by PMU; the injected power of each node in the distribution network and the power of each branch are measured by SCADA; In step 2, the voltage of each node and the current of each branch measured by the PMU are measured and transformed, specifically: in, is the PMU measurement at node i Phase voltage amplitude; is the PMU measurement at node i The voltage phase angle of the phase; The branch ij measured by the PMU Phase current amplitude; The branch ij measured by the PMU The current phase angle of the phase; The injected power of each node and the power of each branch measured by SCADA are equivalently converted as follows: in, is the node i of SCADA measurement The injected active power of the phase; is the node i of SCADA measurement The injected reactive power of the phase; The branch ij measured by SCADA Active power of each phase; The branch ij measured by SCADA Reactive power of each phase; Except for the pseudo-measurement data in interval form, the rest of the measurement data β is converted to interval form season 4. The method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive according to claim 1 is characterized in that: The step 3 is specifically as follows: Mathematical model for three-phase unbalanced interval state estimation of distribution network: [z]=[H][x]+[ε] Where [H] is the Jacobian matrix in interval form; [ε] is the error term in interval form; The weighted least squares method is used to introduce the weight matrix W and the transition phase [y] = [H] [x] - [z], and the mathematical model of the three-phase unbalanced interval state estimation of the distribution network is converted into: [A][X]=[B] Among them, I is the unit matrix; The Krawczyk algorithm is used to iteratively obtain the nonlinear interval state variables: [X k+1 ]=[K(x)]∩[X k ] [K(x)]=C[B]+(IC[A]) Wherein, Mid[A] means taking the middle value of the interval elements in the matrix [A]; k is the number of iterations; n is the number of nodes in the distribution network; m is the measurable quantity; when And|| X k+1 |-| X k ||≤ε, or when the number of iterations k reaches the preset upper limit of the number of iterations, stop the iteration and let Then the interval state estimation vector [x] is obtained.

5. The method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive according to claim 4 is characterized by: When calculating C[B]+(IC[A]), an affine operation is introduced to reduce interval expansion and the interval matrices [B] and [A] are converted into affine matrices. calculate After the calculation is completed, Convert to interval number [K(x)], and then solve [X k+1 ]; The conversion relationship between interval numbers and affine numbers is: Define the number of intervals Affine number ε i =[-1,1] indicates the noise source; Convert interval numbers to affine numbers: Convert affine numbers to interval numbers:

6. The method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive according to claim 4 is characterized by: The solution method of the interval form Jacobian matrix [H] is: Among them, R, S, and T are the node voltage measurement set, node current measurement set, and branch current measurement set in the distribution network respectively; i,j=1,2,...,n; Generate the three-phase admittance matrix Y of the distribution network B : Among them, Y ij represents the admittance matrix of branch ij; is the mutual admittance of the three-phase branch ij; is the self-admittance of the three-phase branch ij; is the mutual conductance between the three phases of branch ij; It is the self-conduction between the three phases of branch ij; is the mutual acceptance between the three phases of branch ij; is the self-susceptance between the three phases of branch ij; The functional relationship between the amount of node injection current measurement and the state quantity is: Where Φ = {a, b, c}; Ω is the set of nodes connected to node i; and They are respectively Conductance and susceptance relative to the φ phase; The functional relationship between the branch current measurement and the state quantity is: Based on the above two sets of functional relationships, the elements in the Jacobian matrix [H] are solved.

7. The method for estimating interval state of a three-phase unbalanced distribution network with digital-analog dual drive according to claim 1 is characterized by: In step 4, the neural network is trained and the loss function used is: Loss = α Loss1 + (1-α) Loss2 Among them, α is a hyperparameter that adjusts the two loss functions; Among them, x is the input of the neural network, y is the output of the neural network; δ is the parameter of the loss function; For the loss function in the training process, physical constraints are introduced to prevent the training results from deviating from the normal range; the per-unit values ​​of the real and imaginary parts of the voltage at each node in the distribution network are in the interval [V re,low ,V re,up ]、[V im,low ,V im,up ] remain stable, and this interval is determined according to the specific distribution network system; in, are the upper and lower bounds of the interval estimator of the real part of the voltage at node i, respectively; are the upper and lower bounds of the interval estimator of the imaginary part of the voltage at node i, respectively.

8. A computer device / equipment / system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images