Power distribution network bad data identification state estimation method based on digital twinning

Abstract

The invention discloses a power distribution network bad data identification state estimation method based on digital twinning, and relates to the technical field of power system state estimation, and the method comprises the steps: collecting real-time measurement data based on a physical power distribution network, and transmitting the real-time measurement data to a digital twinning server through a communication protocol; identifying and correcting bad measurement data in a digital twin database of the server, and complementing the removed measurement data through a BILSTM neural network; and fusing the multi-source measurement data in the aspects of data volume and time scale, and performing linear state estimation to reflect the real-time state of the power distribution network. According to the method, bad data are identified and corrected through the BILSTM neural network, and the data integrity is ensured; by adopting the linear state estimation method based on multi-source data fusion, the problem of time scale difference is solved, accurate and real-time power distribution network perception is realized, and better effects are achieved in the aspects of bad data identification and correction of the power distribution network and the precision and efficiency of state estimation.

Description

Technical Field

[0001] The present invention relates to the technical field of power system state estimation, and in particular to a distribution network bad data identification state estimation method based on digital twins. Background Art

[0002] With the transformation of traditional power grids to smart grids and the continuous development of distributed energy, traditional distribution networks are gradually transforming into active distribution networks with bidirectional power flows, and the operating status and control mode are becoming increasingly complex; the high penetration rate and output uncertainty of new energy will affect the stable operation of the power system and increase the safety risks of the power system; in this case, it is particularly important to build a digital distribution network and realize real-time distribution network status perception.

[0003] State estimation is one of the methods to achieve real-time perception of distribution networks. The complete distribution network state can be obtained through limited measurements, and can be used as a method for digital twin power grids to achieve real-time state perception. However, the accuracy of distribution network state estimation is heavily dependent on the accuracy of measurement data. In actual distribution network measurement data, there are often erroneous measurement data, which seriously affects the accuracy of state estimation. In addition, with the introduction of synchronous phasor measurement devices, there are multi-source measurement data with different measurement data volumes and time scales in the distribution network. Fully integrating multi-source measurement data and performing fast and efficient state estimation are important prerequisites for efficient operation decision-making and dispatching control of distribution networks.

[0004] Therefore, it is necessary to develop new methods to identify and correct bad measurement data, unify multi-source measurements in time and data volume, and perform linear state estimation to accurately and quickly perceive the real-time status of the distribution network. Summary of the invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is: the existing distribution network state estimation technology has the problems of low state estimation accuracy and low computational efficiency when facing multi-source measurement data, bad data and measurement data at different time scales, as well as how to effectively integrate multi-source measurement data.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a method for identifying and estimating the state of distribution network bad data based on digital twins, comprising collecting real-time measurement data based on the physical distribution network, and sending it to the digital twin server through a communication protocol; identifying and correcting the bad measurement data in the digital twin database of the server, and completing the eliminated measurement data through a BILSTM neural network; fusing multi-source measurement data in terms of data volume and time scale, and performing linear state estimation to reflect the real-time state of the distribution network.

[0008] As a preferred solution of the distribution network bad data identification state estimation method based on digital twin described in the present invention, wherein: the collection of real-time measurement data includes transmitting the data of each node collected by the multi-source measurement device of the distribution network to the server through the communication protocol and performing digital twin distribution network modeling.

[0009] As a preferred solution of the distribution network bad data identification state estimation method based on digital twin described in the present invention, wherein: the collection of real-time measurement data also includes obtaining the node voltage amplitude and phase angle, branch current amplitude and phase angle through the PMU measurement device; obtaining the branch active power and reactive power through the SCADA measurement device; obtaining the node injected active power and reactive power through the AMI measurement device.

[0010] As a preferred solution of the method for identifying and estimating the state of distribution network bad data based on digital twins described in the present invention, the identification and correction of bad measurement data includes extracting a historical measurement phase with a duration of T′-1 before the current moment based on the digital twin database, and comparing it with the real-time measurement vector z at the current moment. k+1 The measurement time series [z k+2-T′ ,z k+3-T′ ,···,z k+1 ], calculate the Pearson correlation coefficient of the measurement time series with a duration of T′, and establish a Pearson correlation coefficient matrix based on the correlation coefficient of every two measurements, expressed as:

[0011]

[0012] Among them, ρ ab is an element in the correlation coefficient matrix, representing the temporal correlation between measurement a and measurement b, |ρ ab |<1, the matrix diagonal elements are 1, is the average value of the time series measurement series, z k+2-t is the quantity measurement at time k+2-t, k is the number of the measurement sequence at time k, and t is the time point number; set is the average value of the correlation coefficient of the measurement, expressed as:

[0013]

[0014] Where N is the number of all measurements; based on the bad data generated in the historical database Minimum value of Compared with the case without bad data The maximum value of When bad data is generated Greater than when no bad data is present Setting Thresholds like It is judged as bad data and is eliminated.

[0015] As a preferred solution of the method for identifying and estimating the state of distribution network bad data based on digital twins described in the present invention, the measured data complemented and eliminated by the BILSTM neural network includes a BILSTM body including two independent LSTM structures, the same input sequence is respectively connected to the two forward and backward LSTMs, the hidden layers of the two networks are connected together, and are connected to the output layer for prediction; based on the BILSTM hidden layer, the forward parameter h is saved t With the reverse parameter h k , design the final output of BILSTM t , expressed as:

[0016]

[0017] Among them, y t is the output value at time t, e t is the input value at time t, g and L are sigmoid activation functions, K, K′, W, W′, U, U′ are weight values ​​corresponding to the input value, and h t-1 h t Left side parameter, h k-1 h k The parameters on the right side, D is the neural network parameter after forward LSTM processing, and D′ is the neural network parameter after reverse LSTM processing.

[0018] As a preferred solution of the method for identifying and estimating the state of distribution network bad data based on digital twins described in the present invention, the fusion of multi-source measurement data in terms of data volume and time scale includes unifying the measurement data of PMU, SCADA, and AMI in terms of time scale, using PMU measurement data as the time scale, using linear extrapolation interpolation to construct AMI pseudo-measurement data, and designing a linear extrapolation method, which is expressed as:

[0019]

[0020] Among them, z δ,k AMI measurement data at t l and t l+1 Any time t between k The quantity measurement, T′ δ is the measurement period of AMI measurement data, z δ,l t l The quantity measurement of AMI measurement data at the moment, z δ,l-1 t l-1 The quantity of AMI measurement data at time t, δ is the number of the measurement cycle, and l is the identifier of the measurement time.

[0021] As a preferred solution of the method for identifying state estimation of distribution network bad data based on digital twins described in the present invention, the linear state estimation includes linearly transforming different measurement data, uniformly converting them into equivalent linear quantity measurements with separated imaginary and real parts, and performing linear transformation based on the node voltage phasor and branch current phasor data collected by PMU measurement, which is expressed as:

[0022]

[0023] Among them, i and j are the numbers of the first and last nodes of the line respectively, and p is the phase. are the real and imaginary parts of the equivalent branch current, are the current amplitude and phase angle of the PMU measurement branch ij, are the real and imaginary parts of the voltage at the equivalent node i, respectively. are the voltage amplitude and phase angle of node i measured by PMU, P is PMU measurement, re is the real part, and im is the imaginary part; the branch power measured by SCADA and AMI is measured and converted with the node injection power, which can be expressed as:

[0024]

[0025] in, They are the active power and reactive power of branch ij measured by SCADA respectively; and are the real and imaginary parts of the node voltage obtained in each iteration, Inject active power and reactive power into the nodes collected by AMI respectively. are the real and imaginary parts of the equivalent node injected current, S is SCADA measurement, and A is AMI measurement; the voltage and current covariance of the equivalent measurement is evaluated based on the error propagation law, and the equivalent node voltage covariance matrix It is expressed as:

[0026]

[0027] in, are the autocovariance of the real and imaginary parts of the equivalent node voltage, is the cross-covariance of the real and imaginary parts of the equivalent node voltage, σ U , σ θ is the standard deviation of the node voltage amplitude and phase angle measurement, U i ,θ i is the true value of the node voltage amplitude and phase angle measurement; the covariance matrix of the equivalent branch current converted from the branch power measurement It is expressed as:

[0028]

[0029] in, and are the autocovariance of the real and imaginary parts of the equivalent branch current, is the cross-covariance of the real and imaginary parts of the equivalent branch current, is the standard deviation of branch active power and reactive power, V i-re 、V i-im is the real and imaginary part of the node voltage during the iteration process; based on the measurement transformation, the state estimation measurement equation is constructed, which is expressed as:

[0030] z=h(x)+v

[0031] Wherein, z is the measurement vector, h(x) is the measurement vector function vector, and v is the error vector between z and h(x);

[0032] The state variable x is set to the real and imaginary parts of the voltages of the n three-phase nodes of the distribution network, expressed as:

[0033] x=[V 1,re ,V 1,im ,…,V n,re ,V n,im ] T

[0034] Where T is the transposition operation; based on the transformed measurement data, the measurement vector z is expressed as:

[0035]

[0036] Among them, Z is the different quantities contained in the measurement vector, and are the real and imaginary parts of the equivalent node voltage measurement after transformation, The real and imaginary parts of the current measurement injected into the equivalent node after transformation, is the real part and imaginary part of the equivalent branch current measurement after transformation; a state estimation model is established, and a state estimation mathematical model with equality constraints is established using the weighted least squares method, which is expressed as:

[0037]

[0038] Among them, stc(x) is the zero injection node constraint function, J(x) is the objective function of the least squares estimation, and R is the transformed error covariance; based on the transformed measurement data, the measurement vector function vector h(x) is expressed as:

[0039]

[0040] Among them, ω is the phase accumulation mark in the accumulation formula, including phase a, phase b, and phase c of the distribution network, and Ω is the set of nodes in the entire distribution network. is the corresponding conductance element in the admittance matrix of the distribution network node, is the corresponding susceptance element in the admittance matrix of the distribution network node, G and B are conductance and susceptance respectively; based on the derivative of the measurement function vector h(x), the Jacobian matrix H is calculated, which is expressed as:

[0041]

[0042] Among them, V re 、V im are the real and imaginary parts of the node voltage respectively; the Lagrange multiplier method is used to solve the Jacobian matrix and construct the Lagrange extreme value function, which is expressed as:

[0043]

[0044] Among them, L(x,λ) is the objective function; the partial derivative of the state variable x and the constraint phasor λ is expressed as:

[0045]

[0046] Where C is the zero injection measurement constraint Jacobian matrix, R M is the covariance matrix of the fusion after transformation;

[0047] Based on Newton's method, it is iterated and converted into the least squares estimation equation with equality constraints, which is expressed as:

[0048]

[0049] Among them, o is the number of iterations, x o is the state variable at the oth iteration, λ o is the Lagrange multiplier at the oth iteration, Δλ o is the correction value of the Lagrange multiplier at the oth iteration; the state variables are constrained, expressed as:

[0050] |x o+1 -x o | max <ε

[0051] Among them, ε is the set state estimation accuracy threshold; if the state variable satisfies the constraint, x o+1 The value of is the final estimated value, and the iteration ends.

[0052] Another object of the present invention is to provide a distribution network bad data identification state estimation system based on digital twins, which can solve the problems of low accuracy and poor timeliness of current distribution network state perception technology through a linear state estimation method including bad data identification and correction.

[0053] As a preferred solution of the digital twin-based distribution network bad data identification state estimation system described in the present invention, it includes: a data measurement and communication module, a bad data identification and correction module, and a data fusion linear state estimation module; the data measurement and communication module is used to collect real-time measurement data based on the physical distribution network, and send it to the digital twin server through a communication protocol; the bad data identification and correction module is used to identify and correct the bad measurement data in the digital twin database of the server, and complete the eliminated measurement data through the BILSTM neural network; the data fusion linear state estimation module is used to fuse multi-source measurement data in terms of data volume and time scale, and perform linear state estimation to reflect the real-time state of the distribution network.

[0054] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a method for estimating the state of distribution network bad data identification based on digital twins.

[0055] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for estimating the state of distribution network bad data identification based on digital twins.

[0056] Beneficial effects of the present invention: The distribution network bad data identification state estimation method based on digital twin provided by the present invention identifies bad data in the distribution network measurement data by considering the time series correlation of the distribution network measurement data, thereby improving the robustness of the distribution network state estimation; through BILSTM neural network training, the current moment measurement data is predicted to complete the eliminated data, thereby ensuring the integrity of the measurement data; through the linear state estimation method of multi-source data fusion, the present invention solves the problem of cross-sectional data alignment of measurement data at different time scales, and can realize accurate and real-time distribution network perception, and provides an efficient method for accurate modeling of digital twin distribution networks. The present invention achieves better results in the identification and correction of bad data in the distribution network and the accuracy and efficiency of state estimation, and can be used as a method for establishing a real-time state perception model of a digital twin distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0058] Figure 1 An overall flow chart of a method for identifying and estimating the state of distribution network bad data based on digital twins provided in the first embodiment of the present invention;

[0059] Figure 2 A BILSTM network structure diagram of a distribution network bad data identification state estimation method based on digital twins provided in the first embodiment of the present invention;

[0060] Figure 3 A schematic diagram of linear extrapolation of AMI measurement for a distribution network bad data identification state estimation method based on digital twins provided in the first embodiment of the present invention;

[0061] Figure 4 A flowchart of a multi-source measurement fusion linear state estimation method for a distribution network bad data identification state estimation method based on digital twins provided in the first embodiment of the present invention;

[0062] Figure 5 An IEEE33-node distribution network topology diagram of a distribution network bad data identification state estimation method based on digital twins provided in the second embodiment of the present invention;

[0063] Figure 6 A BILSTM branch active power measurement prediction effect diagram of a distribution network bad data identification state estimation method based on digital twin provided in the second embodiment of the present invention;

[0064] Figure 7 A BILSTM branch reactive power measurement prediction effect diagram of a distribution network bad data identification state estimation method based on digital twin provided in the second embodiment of the present invention;

[0065] Figure 8 A comparison diagram of the actual and estimated values ​​of voltage amplitudes of 33 nodes of a distribution network bad data identification state estimation method based on digital twins provided in the second embodiment of the present invention;

[0066] Fig. 9 A comparison diagram of the true and estimated values ​​of the voltage phase angles of 33 nodes of a distribution network bad data identification state estimation method based on digital twins provided in the second embodiment of the present invention;

[0067] Fig.10A comparison diagram of the actual and estimated voltage amplitudes of nodes 10 at 500 time sections of a distribution network bad data identification state estimation method based on digital twins provided in the second embodiment of the present invention;

[0068] Fig.11 A comparison diagram of the actual value and estimated value of the voltage phase angle of node 10 at 500 time sections of a distribution network bad data identification state estimation method based on digital twin provided in the second embodiment of the present invention;

[0069] Fig.12 A module schematic diagram of a digital twin distribution network bad data identification state estimation system provided for the third embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0071] Example 1, reference Figure 1-Figure 4 , is an embodiment of the present invention, and provides a distribution network bad data identification state estimation method based on digital twin, comprising:

[0072] S1: Collect real-time measurement data based on the physical distribution network and send it to the digital twin server through the communication protocol.

[0073] Furthermore, collecting real-time measurement data includes transmitting the data of each node collected by the multi-source measurement device of the distribution network to the server through the communication protocol and performing digital twin distribution network modeling.

[0074] It should be noted that collecting real-time measurement data also includes obtaining node voltage amplitude and phase angle, branch current amplitude and phase angle through PMU measurement devices; obtaining branch active power and reactive power through SCADA measurement devices; and obtaining node injected active power and reactive power through AMI measurement devices.

[0075] It should also be noted that different types of measuring devices have different data collection cycles. The PMU data collection cycle is in milliseconds, the SCADA data collection cycle is in seconds, and the AMI data collection cycle is in minutes. The digital twin database generated by digital twin distribution network modeling has the function of identifying and correcting bad data, ensuring the accuracy and completeness of the measurement data, thereby improving the speed and accuracy of subsequent state estimation.

[0076] S2: Identify and correct bad measurement data in the server’s digital twin database, and use the BILSTM neural network to complete the eliminated measurement data.

[0077] Furthermore, the identification and correction of bad measurement data includes extracting a historical measurement phase with a duration of T′-1 before the current moment based on the digital twin database and comparing it with the real-time measurement vector z at the current moment. k+1 The measurement time series [z k+2-T′ ,z k+3-T′ ,···,z k+1 ], calculate the Pearson correlation coefficient of the measurement time series with a duration of T′, and establish a Pearson correlation coefficient matrix based on the correlation coefficient of every two measurements, expressed as:

[0078]

[0079] Among them, ρ ab is an element in the correlation coefficient matrix, representing the temporal correlation between measurement a and measurement b, |ρ ab |<1, the matrix diagonal elements are 1, is the average value of the time series measurement series, z k+2-t is the quantity measurement at time k+2-t, k is the number of the measurement sequence at time k, and t is the time point number; set is the average value of the correlation coefficient of the measurement, expressed as:

[0080]

[0081] Where N is the number of all measurements; based on the bad data generated in the historical database Minimum value of Compared with the case without bad data The maximum value of When bad data is generated Greater than when no bad data is present Setting Thresholds like It is judged as bad data and is eliminated.

[0082] It should be noted that the measurement data that is eliminated by the BILSTM neural network includes two independent LSTM structures. The same input sequence is connected to the two LSTMs, the forward and backward LSTMs, respectively. The hidden layers of the two networks are connected together and connected to the output layer for prediction. The forward parameter h is saved based on the BILSTM hidden layer. t With the reverse parameter h k , design the final output of BILSTM t , expressed as:

[0083]

[0084] Among them, y t is the output value at time t, e t is the input value at time t, g and L are sigmoid activation functions, K, K′, W, W′, U, U′ are weight values ​​corresponding to the input value, and h t-1 h t Left side parameter, h k-1 h k The parameters on the right side, D is the neural network parameter after forward LSTM processing, and D′ is the neural network parameter after reverse LSTM processing.

[0085] It should also be noted that, constrained by the topological structure of the physical distribution network, there is a temporal correlation between various measurement data in the distribution network in the same area. ab The closer it is to 1, the stronger the correlation between the two corresponding measurement data is. ab The closer it is to 0, the weaker the correlation between the two measured data. After the bad data is eliminated, the missing data is supplemented based on the BILSTM neural network. The LSTM network is a neural network with the ability to remember long-term and short-term information. By introducing a gating mechanism to control the storage and forgetting of features, it gains the ability to increase and delete information transmitted in the network. The BILSTM body includes two independent LSTM structures. The same input sequence is connected to the forward and backward LSTMs respectively, and then the hidden layers of the two networks are connected together and connected to the output layer for prediction. The network structure of BILSTM is as follows: Figure 2 shown.

[0086] S3: Fusion of multi-source measurement data in terms of data volume and time scale, and linear state estimation to reflect the real-time status of the distribution network.

[0087] Furthermore, the fusion of multi-source measurement data in terms of data volume and time scale includes unifying the measurement data of PMU, SCADA, and AMI in terms of time scale. PMU measurement data is used as the time scale, and the linear extrapolation interpolation method is used to construct AMI pseudo-measurement data. The linear extrapolation method is designed, which is expressed as:

[0088]

[0089] Among them, z δ,k AMI measurement data at t l and t l+1 Any time t between k The quantity measurement, T′ δ is the measurement period of AMI measurement data, zδ,l t l The quantity measurement of AMI measurement data at the moment, z δ,l-1 t l-1 The quantity of AMI measurement data at time t, δ is the number of the measurement cycle, and l is the identifier of the measurement time.

[0090] It should be noted that linear state estimation includes linear transformation of different measurement data, uniformly converting them into equivalent linear quantity measurements with separated imaginary and real parts, and performing linear transformation based on the node voltage phasor and branch current phasor data collected by PMU measurement, which can be expressed as:

[0091]

[0092] Among them, i and j are the numbers of the first and last nodes of the line respectively, and p is the phase. are the real and imaginary parts of the equivalent branch current, are the current amplitude and phase angle of the PMU measurement branch ij, are the real and imaginary parts of the voltage at the equivalent node i, respectively. are the voltage amplitude and phase angle of node i measured by PMU, P is PMU measurement, re is the real part, and im is the imaginary part; the branch power measured by SCADA and AMI is measured and converted with the node injection power, which can be expressed as:

[0093]

[0094] in, They are the active power and reactive power of branch ij measured by SCADA respectively; and are the real and imaginary parts of the node voltage obtained in each iteration, Inject active power and reactive power into the nodes collected by AMI respectively. are the real and imaginary parts of the equivalent node injected current, S is SCADA measurement, and A is AMI measurement; the voltage and current covariance of the equivalent measurement is evaluated based on the error propagation law, and the equivalent node voltage covariance matrix It is expressed as:

[0095]

[0096] in, are the autocovariance of the real and imaginary parts of the equivalent node voltage, is the cross-covariance of the real and imaginary parts of the equivalent node voltage, σ U , σ θ is the standard deviation of the node voltage amplitude and phase angle measurement, U i ,θ iis the true value of the node voltage amplitude and phase angle measurement; the covariance matrix of the equivalent branch current converted from the branch power measurement It is expressed as:

[0097]

[0098] in, and are the autocovariance of the real and imaginary parts of the equivalent branch current, is the cross-covariance of the real and imaginary parts of the equivalent branch current, is the standard deviation of branch active power and reactive power, V i-re 、V i-im is the real and imaginary part of the node voltage during the iteration process; based on the measurement transformation, the state estimation measurement equation is constructed, which is expressed as:

[0099] z=h(x)+v

[0100] Where z is the measurement vector, h(x) is the measurement vector function vector, and v is the error vector between z and h(x). The state variable x is set to the real and imaginary parts of the voltages of the n three-phase nodes of the distribution network, expressed as:

[0101] x=[V 1,re ,V 1,im ,…,V n,re ,V n,im ] T

[0102] Where T is the transposition operation; based on the transformed measurement data, the measurement vector z is expressed as:

[0103]

[0104] Among them, Z is the different quantities contained in the measurement vector, and are the real and imaginary parts of the equivalent node voltage measurement after transformation, The real and imaginary parts of the current measurement injected into the equivalent node after transformation, is the real part and imaginary part of the equivalent branch current measurement after transformation; a state estimation model is established, and a state estimation mathematical model with equality constraints is established using the weighted least squares method, which is expressed as:

[0105]

[0106] Among them, stc(x) is the zero injection node constraint function, J(x) is the objective function of the least squares estimation, and R is the transformed error covariance; based on the transformed measurement data, the measurement vector function vector h(x) is expressed as:

[0107]

[0108] Among them, ω is the phase accumulation mark in the accumulation formula, including phase a, phase b, and phase c of the distribution network, and Ω is the set of nodes in the entire distribution network. is the corresponding conductance element in the admittance matrix of the distribution network node, is the corresponding susceptance element in the admittance matrix of the distribution network node, G and B are conductance and susceptance respectively; based on the derivative of the measurement function vector h(x), the Jacobian matrix H is calculated, which is expressed as:

[0109]

[0110] Among them, V re 、V im are the real and imaginary parts of the node voltage respectively; the Lagrange multiplier method is used to solve the Jacobian matrix and construct the Lagrange extreme value function, which is expressed as:

[0111]

[0112] Among them, L(x,λ) is the objective function; the partial derivative of the state variable x and the constraint phasor λ is expressed as:

[0113]

[0114] Where C is the zero injection measurement constraint Jacobian matrix, R M is the covariance matrix of the fusion after transformation.

[0115] Based on Newton's method, it is iterated and converted into the least squares estimation equation with equality constraints, which is expressed as:

[0116]

[0117] Among them, o is the number of iterations, x o is the state variable at the oth iteration, λ o is the Lagrange multiplier at the oth iteration, Δλ o is the correction value of the Lagrange multiplier at the oth iteration; the state variables are constrained, expressed as:

[0118] |x o+1 -x o | max <ε

[0119] Among them, ε is the set state estimation accuracy threshold; if the state variable satisfies the constraint, x o+1 The value of is the final estimated value, and the iteration ends.

[0120] It should also be noted that the three types of measurement data are unified in time scale. For PMU measurement, because its measurement sampling period is small, the data accuracy is high, and all measurements have accurate time scales, PMU data is selected as the benchmark; SCADA measurement data does not have a unified time scale. When PMU measurement and SCADA measurement are used for state estimation at the same time, data alignment and time delay need to be considered. The present invention directly uses the measurement data sampled most recently at the state estimation time as the measurement value after data fusion; the sampling time interval of AMI measurement is relatively large, at the minute level, but there is a unified time scale, and linear extrapolation interpolation method is used to form AMI pseudo-measurement data. Figure 3 This is a schematic diagram of linear extrapolation of AMI measurement. Figure 3 It can be seen that the data between z2 and z3 can be derived using the data at time z1 and z2; Figure 4 This is a flow chart of the multi-source measurement fusion linear state estimation method. The process is to first input the corrected data and the distribution network topology data, perform time scale fusion of the measurement data, perform linear transformation of the measurement data, unify it into branch current measurement, and finally establish a state estimation model. Considering the zero injection node constraint, the state variable is assigned an initial value, the number of iterations is set to o = 0, and the number of iterations is increased by 1 each time. The iterative solution is obtained by x o+1 If the convergence condition is met, the distribution network state quantity is output; if the convergence condition is not met, the iterative solution is continued until the convergence condition is met and the distribution network state quantity is output.

[0121] Example 2, reference Figure 5-Figure 11 , is an embodiment of the present invention, and provides a distribution network bad data identification state estimation method based on digital twin. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0122] In order to verify the distribution network bad data identification state estimation method based on digital twin, the present invention builds an IEEE33-node distribution network model on RTDS (Real Time Digital Simulator) as the basis for simulation. The IEEE33-node distribution network topology diagram is shown in Figure 5 As shown, real fluctuating load data are added to nodes 18 and 22; PMU measurement devices are installed on nodes 1, 3, 6, 13, 21, and 28, and the measurement standard deviation is set to 0.005; AMI measurement devices are installed on nodes 2, 7, 18, 20, 27, and 33, and the measurement standard deviation is set to 0.005; SCADA measurement is installed on the remaining nodes, and the measurement standard deviation is set to 0.01.

[0123] In order to verify the effect of bad data identification of distribution network measurement, 3000 continuous section measurement data of distribution network measurement are selected, 5% bad data are randomly added to the first 1000 sections, 10% bad data are randomly added to the middle 1000 sections, and 15% bad data are randomly added to the last 1000 sections. The bad data are set as follows: normal voltage measurement increases or decreases by 20%~30%, and normal power measurement increases or decreases by 50%~60%. The bad data identification effect is shown in Table 1.

[0124] Table 1 Bad data identification effect

[0125] Bad data ratio Recognition rate False positive rate 5％ 97.6％ 2.6％ 10％ 95.2％ 3.2％ 15％ 92.8％ 4.1％

[0126] It can be seen from the data in Table 1 that the bad data identification method of the present invention has a higher recognition rate and a lower misjudgment rate, and can still maintain a good recognition effect when the proportion of bad data increases, and can accurately identify bad data.

[0127] To verify the prediction effect of the measurement data based on BILSTM, the active power and reactive power measurement prediction of branch 3-4 is taken as an example. Figure 6 This is the comparison between the actual measured value and the predicted value of the active power of branch 3-4. Figure 7 This is the comparison between the actual measured value and the predicted value of reactive power in branch 3-4. Figure 6 , Figure 7 It can be seen that when the measurement data fluctuates, the prediction method proposed in the present invention has a very high degree of fit with the true value. The proposed BILSTM neural network can comprehensively capture and understand the context and dependencies in the time series data, has high prediction accuracy, and can be used as a method to complete the measurement data after removing bad data.

[0128] In order to verify the accuracy of distribution network state estimation, the state estimation of the 33-node distribution network is carried out using the corrected and fused measurements. Figure 8 Comparison between the actual and estimated voltage amplitudes of 33 nodes at a certain time section. Fig. 9 Comparison between the actual and estimated voltage phase angles of 33 nodes at a certain time section. Fig.10 Comparison between the actual and estimated voltage amplitudes of node 10 at 500 time sections. Fig.11 The comparison between the true value and the estimated value of the voltage phase angle of node 10 at 500 time sections shows that the error between the estimated value and the true value of node 33 is very small, and it still has a high accuracy when there is load fluctuation.

[0129] The simulation results show that the proposed distribution network bad data identification state estimation method based on digital twin can realize efficient and stable operation of the device, has strong anti-error ability, high accuracy and efficiency, and verifies the effectiveness of the proposed design method in the results. Therefore, our invention is creative.

[0130] Example 3, reference Fig.12 , is an embodiment of the present invention, and provides a distribution network bad data identification and state estimation system based on digital twin, including a data measurement and communication module, a bad data identification and correction module, and a data fusion linear state estimation module.

[0131] The data measurement and communication module is used to collect real-time measurement data based on the physical distribution network and send it to the digital twin server through the communication protocol; the bad data identification and correction module is used to identify and correct the bad measurement data in the digital twin database of the server, and complete the eliminated measurement data through the BILSTM neural network; the data fusion linear state estimation module is used to fuse multi-source measurement data in terms of data volume and time scale, and perform linear state estimation to reflect the real-time state of the distribution network.

[0132] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0133] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0134] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0135] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A distribution network bad data identification state estimation method based on digital twin, characterized in that: include: Collect real-time measurement data based on the physical distribution network and send it to the digital twin server through the communication protocol; Identify and correct bad measurement data in the server’s digital twin database, and use the BILSTM neural network to complete the eliminated measurement data; The multi-source measurement data are fused in terms of data volume and time scale, and linear state estimation is performed to reflect the real-time status of the distribution network.

2. The method for identifying and estimating the state of distribution network bad data based on digital twins according to claim 1, characterized in that: The real-time measurement data collection includes transmitting the data of each node collected by the multi-source measurement device of the distribution network to the server through the communication protocol and performing digital twin distribution network modeling.

3. The method for identifying and estimating the state of distribution network bad data based on digital twins according to claim 2, characterized in that: The collecting of real-time measurement data also includes obtaining node voltage amplitude and phase angle, branch current amplitude and phase angle through a PMU measurement device; Obtain branch active power and reactive power through SCADA measurement devices; The active power and reactive power injected into the node are obtained through the AMI measurement device.

4. The method for identifying and estimating the state of distribution network bad data based on digital twins according to claim 3, characterized in that: The identification and correction of bad measurement data includes extracting a historical measurement phase with a duration of T′-1 before the current moment based on the digital twin database and comparing it with the real-time measurement vector z at the current moment. k+1 The measurement time series [z k+2-T ,,z k+3-T ,,·…,z k+1 ], calculate the Pearson correlation coefficient of the measurement time series with a duration of T′, and establish a Pearson correlation coefficient matrix based on the correlation coefficient of every two measurements, expressed as: Among them, ρ ab is an element in the correlation coefficient matrix, representing the temporal correlation between measurement a and measurement b, |ρ ab |<1, the matrix diagonal elements are 1, is the average value of the time series measurement series, z k+2-t is the quantity measurement at time k+2-t, k is the number of the measurement sequence at time k, and t is the time point number; set up is the average value of the correlation coefficient of the measurement, expressed as: Where N is the number of all quantity measurements; Based on the bad data generated in the historical database Minimum value of Compared with the case without bad data The maximum value of When bad data is generated Greater than when no bad data is present Setting Thresholds like It is judged as bad data and is eliminated.

5. The method for identifying and estimating the state of distribution network bad data based on digital twins according to claim 4, characterized in that: The measurement data that is complemented and eliminated by the BILSTM neural network includes a BILSTM main body including two independent LSTM structures, the same input sequence is respectively connected to the two forward and backward LSTMs, the hidden layers of the two networks are connected together, and are connected to the output layer for prediction; Save the forward parameter h based on the BILSTM hidden layer t With the reverse parameter h k , design the final output of BILSTM t , expressed as: Among them, y t is the output value at time t, e t is the input value at time t, g and L are sigmoid activation functions, K, K′, W, W′, U, U′ are weight values ​​corresponding to the input value, and h t-1 h t Left side parameter, h k-1 h k The parameters on the right side, D is the neural network parameter after forward LSTM processing, and D′ is the neural network parameter after reverse LSTM processing.

6. The method for identifying and estimating the state of distribution network bad data based on digital twins according to claim 5, characterized in that: The fusion of multi-source measurement data in terms of data volume and time scale includes unifying the measurement data of PMU, SCADA, and AMI in terms of time scale, using PMU measurement data as the time scale, using linear extrapolation interpolation to construct AMI pseudo measurement data, and designing a linear extrapolation method, which is expressed as: Among them, z δ，k AMI measurement data at t l and t l+1 Any time t between k The quantity measurement, T′ δ is the measurement period of AMI measurement data, z δ，l t l The quantity measurement of AMI measurement data at the moment, z δ，l-1 t l-1 The quantity of AMI measurement data at time t, δ is the number of the measurement cycle, and l is the identifier of the measurement time.

7. The method for identifying and estimating the state of distribution network bad data based on digital twins according to claim 6, characterized in that: The linear state estimation includes linear transformation of different measurement data, uniformly converting them into equivalent linear quantity measurements with separated imaginary and real parts, and performing linear transformation based on the node voltage phasor and branch current phasor data collected by PMU measurement, which can be expressed as: Among them, i and j are the numbers of the first and last nodes of the line respectively, and p is the phase. are the real and imaginary parts of the equivalent branch current, are the current amplitude and phase angle of the PMU measurement branch ij, are the real and imaginary parts of the voltage at the equivalent node i, respectively. are the voltage amplitude and phase angle of node i measured by PMU, P is PMU measurement, re is the real part, and im is the imaginary part; The branch power measured by SCADA and AMI is measured and converted to the node injection power, which can be expressed as: in, They are the active power and reactive power of branch ij measured by SCADA respectively; and are the real and imaginary parts of the node voltage obtained in each iteration, Inject active power and reactive power into the nodes collected by AMI respectively. are the real and imaginary parts of the equivalent node injected current, S is the SCADA measurement, and A is the AMI measurement; Evaluate the voltage and current covariance of equivalent measurements based on the error propagation law, and the equivalent node voltage covariance matrix It is expressed as: in, are the autocovariance of the real and imaginary parts of the equivalent node voltage, is the cross-covariance of the real and imaginary parts of the equivalent node voltage, σ U , σ θ is the standard deviation of the node voltage amplitude and phase angle measurement, U i ,θ i It is the true value of node voltage amplitude and phase angle measurement; The covariance matrix of the equivalent branch current converted from branch power measurement It is expressed as: in, and are the autocovariance of the real and imaginary parts of the equivalent branch current, is the cross-covariance of the real and imaginary parts of the equivalent branch current, is the standard deviation of branch active power and reactive power, V i-re 、V i-im are the real and imaginary parts of the node voltage during the iteration process; Based on the measurement transformation, the state estimation measurement equation is constructed, which is expressed as: z=h(x)+v Wherein, z is the measurement vector, h(x) is the measurement vector function vector, and v is the error vector between z and h(x); The state variable x is set to the real and imaginary parts of the voltages of the n three-phase nodes of the distribution network, expressed as: x=[V 1，re ,V 1，im ,...,V n，re ,V n，im ] T Where, T is the transpose operation; Based on the transformed measurement data, the measurement vector z is expressed as: Among them, Z is the different quantities contained in the measurement vector, and are the real and imaginary parts of the equivalent node voltage measurement after transformation, The real and imaginary parts of the current measurement injected into the equivalent node after transformation, are the real and imaginary parts of the equivalent branch current measurement after transformation; A state estimation model is established, and a mathematical model of state estimation with equality constraints is established using the weighted least squares method, which is expressed as: Among them, stc(x) is the zero injection node constraint function, J(x) is the objective function of the least squares estimation, and R is the transformed error covariance; Based on the transformed measurement data, the measurement vector function vector h(x) is expressed as: Among them, ω is the phase accumulation mark in the accumulation formula, including phase a, phase b, and phase c of the distribution network, and Ω is the set of nodes in the entire distribution network. is the corresponding conductance element in the admittance matrix of the distribution network node, is the corresponding susceptance element in the admittance matrix of the distribution network node, G and B are conductance and susceptance respectively; Based on the derivative of the measurement function vector h(x), the Jacobian matrix H is calculated and expressed as: Among them, V re 、V im are the real and imaginary parts of the node voltage respectively; The Lagrange multiplier method is used to solve the Jacobian matrix and construct the Lagrange extreme value function, which is expressed as: Among them, L(x, λ) is the objective function; The partial derivative of the state variable x and the constraint phasor λ is expressed as: Where C is the zero injection measurement constraint Jacobian matrix, R M is the covariance matrix of the fusion after transformation; Based on Newton's method, it is iterated and converted into the least squares estimation equation with equality constraints, which is expressed as: Among them, o is the number of iterations, x o is the state variable at the oth iteration, λ o is the Lagrange multiplier at the oth iteration, Δλ o is the correction value of the Lagrange multiplication at the oth iteration; Constraints on state variables are expressed as: |x o+1 -x o | max <e Among them, ε is the set state estimation accuracy threshold; If the state variables satisfy the constraints, x o+1 The value of is the final estimated value, and the iteration ends.

8. A system using the distribution network bad data identification state estimation method based on digital twin as described in any one of claims 1 to 7, characterized in that: It includes data measurement and communication module, bad data identification and correction module, and data fusion linear state estimation module; The data measurement and communication module is used to collect real-time measurement data based on the physical distribution network and send it to the digital twin server through the communication protocol; The bad data identification and correction module is used to identify and correct bad measurement data in the digital twin database of the server, and to complete the eliminated measurement data through the BILSTM neural network; The data fusion linear state estimation module is used to fuse multi-source measurement data in terms of data volume and time scale, and perform linear state estimation to reflect the real-time state of the distribution network.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the distribution network bad data identification state estimation method based on digital twins described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the distribution network bad data identification state estimation method based on digital twins described in any one of claims 1 to 7 are implemented.

Images