A low voltage distribution network topology search method based on user meter data

Abstract

A low-voltage distribution network topology search method based on user meter data includes the following steps: 1) collecting user meter data of m time sections to determine the number of sections for calculating a topology; 2) randomly selecting K continuous section sample TSSs, and setting the sample ordinal number k=1; 3) calculating the user identification sequence IS under the kth section sample TSS; 4) using the user identification sequence IS to sequentially identify the remaining K-1 section sample TSSs, obtain K-1 possible distribution network topologies, and store the K-1 possible distribution network topologies in a result set R; 5) judging whether k=K is established, if not, setting k=k+1, and returning to step 3), if yes, entering step 6); 6) selecting the final distribution network topology from the result set R by voting. The present invention adds a distribution network topology search method of multiple section intervals on the basis of the topology recognition algorithm, which can reduce the accidental errors of the original calculation algorithm based on a single section interval data, and obtain a relatively accurate and stable topological structure.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network topology identification, and in particular to a low-voltage distribution network topology search method based on user electric meter data. Background Art

[0002] The task of power grid topology identification is to transform the physical model described by nodes into a mathematical model described by buses according to the changes in given switch information, and to transform the original physical nodes into topological nodes after considering the switch status.

[0003] The power grid is divided into transmission network and distribution network. The data collection and monitoring configuration of the transmission network has been quite complete. The dispatchers can easily obtain the topological structure of the transmission network and identify topological errors through methods such as state estimation. The distribution network is usually designed in a closed loop and operated in an open loop. It maintains a radial shape during normal operation. When a fault occurs or optimal control is implemented, the section switches and contact lines are activated, and the topological structure of the distribution network changes, causing the flow direction of some branches to change. Under the current level of distribution automation, there are no distribution automation terminals at some section switches of the distribution network. When these switches are activated, since there is no telesignaling and telemetry function, manual inspection and reporting are required, resulting in the inability to update the network topology stored in the system in a timely manner, thereby affecting the safe and economic operation of the distribution network. At the same time, with the large-scale access of distributed power sources, the traditional unidirectional flow becomes a bidirectional interactive flow. If the flow direction cannot be determined, it is easy to cause the relay protection device to malfunction. A clear topology of the distribution network is the premise and basis for flow calculation, fault location, load management and optimization. At present, in China, the penetration rate of smart meters has reached 100%, which provides a large amount of historical data for research and calculation.

[0004] However, the existing distribution network topology identification algorithm does not include verification of the topology structure. The calculation results are very likely to be accidental due to the randomness of the data used, and the accuracy of the low-voltage topology structure identified by the algorithm cannot be guaranteed. Therefore, it is necessary to consider a topology search method based on this algorithm, based on a large amount of data, so that the final distribution network topology is relatively accurate and reliable. Summary of the invention

[0005] The object of the present invention is to provide a low voltage distribution network topology search method based on user meter data, comprising the following steps:

[0006] 1) Collect user meter data at m time sections and determine the number of sections for calculating the topology once, where m is a positive integer.

[0007] 2) Based on the number of sections calculated once and the total number of time sections m, K consecutive section samples TSS are randomly selected, and the sample ordinal number k=1, where K is a positive integer.

[0008] 3) Use the topology recognition algorithm to calculate the user recognition order IS belonging to the kth section sample TSS.

[0009] 4) Use the user identification sequence IS to identify the remaining K-1 section samples TSS in turn, obtain K-1 possible distribution network topologies, and store the K-1 possible distribution network topologies in the result set R.

[0010] 5) Determine whether k=K holds. If not, set k=k+1 and return to step 3). If so, proceed to step 6).

[0011] 6) Select the final distribution network topology from the result set R by voting.

[0012] Furthermore, the value of the total number K of the cross-section samples TSS is as follows:

[0013] K=a*(m÷TS) (1)

[0014] Where m is the total number of time sections. TS is the number of sections for calculating the topology once. a is the adjustment coefficient, and 1≤a≤2.

[0015] Further, the step of using the topology recognition algorithm to calculate the user identification sequence IS belonging to the kth section sample TSS includes:

[0016] 3.1) Calculate the voltage distance matrix V of the node to be judged dist , and find the voltage distance matrix V dist The node pair (i, j) corresponding to the smallest non-zero value in .

[0017] 3.2) Determine the series-parallel relationship of the node pair (i, j) and whether there is a zero injection sink node, and store the obtained series-parallel relationship of the node pair (i, j) into the series-parallel relationship set T. If there is a zero injection sink node, store the zero injection sink node into the zero injection power node set S.

[0018] 3.3) Based on the series-parallel relationship of the node pair (i, j) and the zero injection confluence node, update the voltage, active power, and reactive power of the node to be judged.

[0019] 3.4) Determine whether the number of updated nodes to be determined is 1. If so, output the zero injection power node set S and the series-parallel relationship set T of all nodes to obtain the user identification sequence IS. If not, return to step 3.1).

[0020] Furthermore, the voltage distance matrix V of the node to be determined dist The elements in Euclidean_distance are as follows:

[0021]

[0022] Where V i 、V j Respectively represent the voltage of node i and node j. t is the time section index, t = 1, 2, ..., m, m is the total number of time sections.

[0023] Furthermore, the step of determining the series-parallel relationship of the node pair (i, j) and whether there is a zero injection confluence node includes:

[0024] 3.2.1) Construct the linear overdetermined equation of the zero injection confluence node and solve it using the least squares method.

[0025] 3.2.2) Based on the solution of the linear overdetermined equation, a nonlinear power flow equation is constructed, and the line impedance is obtained by solving the nonlinear power flow equation through a numerical optimization algorithm.

[0026] 3.2.3) Determine the series-parallel relationship of the node pair (i, j) and the zero injection confluence node based on the solved line impedance.

[0027] Furthermore, the linear overdetermined equation of the zero injection confluence node is as follows:

[0028]

[0029] In the formula, V0, V i 、V j are the voltages of zero injection into the confluence node, node i, and node j, respectively, P i , P j are the active powers of nodes i and j respectively. i , Q j are the reactive power of node i and node j respectively. i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line, respectively. i , X j They are the reactances on the zero-injection bus node and the node i line, and the zero-injection bus node and the node j line, respectively.

[0030] Further, the calculation formula for solving the linear overdetermined equation using the least squares method is as follows:

[0031]

[0032] Where V it 、V jt Respectively represent the voltage of node i and node j in the tth time section. t=1,2...m, m is the total number of time sections. Pit , P jt are the active powers of nodes i and j at the tth time section respectively. it , Q jt are the reactive power of node i and node j at the tth time section respectively. i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line, respectively. i , X j They are the reactances on the zero-injection bus node and the node i line, and the zero-injection bus node and the node j line, respectively.

[0033] Furthermore, the line impedance is as follows:

[0034]

[0035] e t (Z) = Q t -f(P t ,Z) (6)

[0036] Where t is the time section index, m is the total number of time sections, and RSS is the line impedance. t (Z) is the residual of the tth time section.

[0037] Among them, the voltage square difference Q at the tth time section t and the nonlinear power flow equation f(P t ,Z) is as follows:

[0038]

[0039]

[0040] Where V it 、V jt Respectively represent the voltage of node i and node j at the tth time section. it , P jt are the active powers of nodes i and j at the tth time section respectively. it , Q jt are the reactive power of node i and node j at the tth time section respectively. it , R jt are the resistances of the zero-injection confluence node and node i line, and the zero-injection confluence node and node j line at the tth time section. it , X jt They are the reactances on the zero-injection confluence node and the line at node i, and the zero-injection confluence node and the line at node j at the tth time section respectively.

[0041] Further, the conditions for determining the series-parallel relationship of the node pair (i, j) and the zero injection confluence node are as follows:

[0042] If the resistor |R j |>κ|R i |、Reactance|X j |>κ|X i | and resistance |R i |<τ、Reactance|X i |<τ, then node i is the parent node of node j, and there is no zero injection confluence node at this time, where κ and τ are both control parameters.

[0043] If the resistor |R i |>κ|R j |、Reactance|X i |>κ|X j | and resistance |R j |<τ、Reactance|X j |<τ, then node j is the parent node of node i, and there is no zero injection confluence node.

[0044] If neither of the above two conditions is met, then node i and node j are brother nodes, and the two nodes are in a parallel relationship. In this case, node i and node j are merged and regarded as a zero injection confluence node.

[0045] Furthermore, if node i is the parent node of node j, the updated voltage of the node to be determined is the voltage V of node i. i , the updated active power and reactive power of the node to be judged are as follows:

[0046]

[0047] Where V j is the voltage at node j, P i , P j are the active powers of nodes i and j respectively. i , Q j are the reactive power of node i and node j respectively. P' and Q' are the updated active power and reactive power of the node to be judged respectively. j is the line resistance at node j. j is the line reactance at node j.

[0048] If node j is the parent node of node i, the updated voltage of the node to be determined is the voltage V of node j. j , the updated active power and reactive power of the node to be judged are as follows:

[0049]

[0050] Where Vi is the voltage at node i, R i is the line resistance at node i. i is the line reactance of node i.

[0051] If node i and node j are brother nodes, the updated voltage, active power, and reactive power of the node to be judged are as follows:

[0052]

[0053]

[0054] Where V k is the updated voltage of the node to be judged, V k(i) 、V k(j) are respectively the updated components of the voltage of the node to be determined at node i and node j.

[0055] The technical effect of the present invention is unquestionable. The present invention is based on a topological identification algorithm for a topological distribution network from top to bottom. On the basis of this algorithm, the characteristics of this algorithm (such as determining the node connection relationship by solving nonlinear overdetermined equations; user identification order is the key to topological identification) are utilized to add a topological search method, which includes a cross-validation and voting process. By selecting different time section intervals to repeat the calculation, the most accurate topology can be obtained as much as possible, thereby improving the accuracy of topological identification. Experiments using field data have shown that the topological search method can indeed improve the accuracy of identification.

[0056] The present invention adds a distribution network topology search method for multiple section intervals on the basis of the existing topology recognition algorithm. This method can reduce the accidental errors of the original calculation algorithm based on single section interval data and obtain a relatively accurate and stable topology structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is a flow chart of the present invention;

[0058] Figure 2 It is a flow chart of the low voltage distribution network topology identification algorithm in the present invention;

[0059] Figure 3 It is a schematic diagram of voltage drop in the low voltage distribution network topology identification algorithm;

[0060] Figure 4 It is the line voltage drop vector diagram in the low voltage distribution network topology identification algorithm. DETAILED DESCRIPTION

[0061] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0062] Embodiment 1:

[0063] See also Figures 1 to 4 , a low voltage distribution network topology search method based on user meter data, comprising the following steps:

[0064] 1) Collect user meter data at m time sections and determine the number of sections for calculating the topology once, where m is a positive integer.

[0065] 2) Based on the number of sections calculated once and the total number of time sections m, K consecutive section samples TSS are randomly selected, and the sample ordinal number k=1, where K is a positive integer.

[0066] 3) Use the topology recognition algorithm to calculate the user recognition order IS belonging to the kth section sample TSS.

[0067] 4) Use the user identification sequence IS to identify the remaining K-1 section samples TSS in turn, obtain K-1 possible distribution network topologies, and store the K-1 possible distribution network topologies in the result set R.

[0068] 5) Determine whether k=K holds. If not, set k=k+1 and return to step 3). If so, proceed to step 6).

[0069] 6) Select the final distribution network topology from the result set R by voting.

[0070] Embodiment 2:

[0071] A low-voltage distribution network topology search method based on user meter data, the main technical content is shown in Example 1, further, the value of the total number K of section samples TSS is as follows:

[0072] K=a*(m÷TS) (1)

[0073] Where m is the total number of time sections. TS is the number of sections for calculating the topology once. a is the adjustment coefficient, and 1≤a≤2.

[0074] Embodiment 3:

[0075] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 2. Further, the step of using a topology recognition algorithm to calculate a user identification sequence IS belonging to the k-th section sample TSS comprises:

[0076] 3.1) Calculate the voltage distance matrix V of the node to be judged dist , and find the voltage distance matrix V dist The node pair (i, j) corresponding to the smallest non-zero value in .

[0077] 3.2) Determine the series-parallel relationship of the node pair (i, j) and whether there is a zero injection sink node, and store the obtained series-parallel relationship of the node pair (i, j) into the series-parallel relationship set T. If there is a zero injection sink node, store the zero injection sink node into the zero injection power node set S.

[0078] 3.3) Based on the series-parallel relationship of the node pair (i, j) and the zero injection confluence node, update the voltage, active power, and reactive power of the node to be judged.

[0079] 3.4) Determine whether the number of updated nodes to be determined is 1. If so, output the zero injection power node set S and the series-parallel relationship set T of all nodes to obtain the user identification sequence IS. If not, return to step 3.1).

[0080] Embodiment 4:

[0081] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 3, further, the voltage distance matrix V of the node to be determined dist The elements in Euclidean_distance are as follows:

[0082]

[0083] Where V i 、V j Respectively represent the voltage of node i and node j. t is the time section index, t = 1, 2, ..., m, m is the total number of time sections.

[0084] Embodiment 5:

[0085] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 4. Further, the step of determining the series-parallel relationship of the node pair (i, j) and whether there is a zero injection confluence node includes:

[0086] 3.2.1) Construct the linear overdetermined equation of the zero injection confluence node and solve it using the least squares method.

[0087] 3.2.2) Based on the solution of the linear overdetermined equation, a nonlinear power flow equation is constructed, and the line impedance is obtained by solving the nonlinear power flow equation through a numerical optimization algorithm.

[0088] 3.2.3) Determine the series-parallel relationship of the node pair (i, j) and the zero injection confluence node based on the solved line impedance.

[0089] Embodiment 6:

[0090] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 5. Further, the linear overdetermined equation of the zero-injection confluence node is as follows:

[0091]

[0092] In the formula, V0, V i 、V j are the voltages of zero injection into the confluence node, node i, and node j, respectively, P i , P j are the active powers of nodes i and j respectively. i , Q j are the reactive power of node i and node j respectively. i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line, respectively. i , X j They are the reactances on the zero-injection bus node and the node i line, and the zero-injection bus node and the node j line, respectively.

[0093] Embodiment 7:

[0094] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 6. Further, the calculation formula for solving the linear overdetermined equation using the least squares method is as follows:

[0095]

[0096] Where V it 、V jt Respectively represent the voltage of node i and node j in the tth time section. t=1,2...m, m is the total number of time sections. P it , P jt are the active powers of nodes i and j at the tth time section respectively. it , Q jt are the reactive power of node i and node j at the tth time section respectively. i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line, respectively. i , X jThey are the reactances on the zero-injection bus node and the node i line, and the zero-injection bus node and the node j line, respectively.

[0097] Embodiment 8:

[0098] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 7, and further, the line impedance is as follows:

[0099]

[0100] e t (Z) = Q t -f(P t ,Z) (6)

[0101] Where t is the time section index, m is the total number of time sections, and RSS is the line impedance. t (Z) is the residual of the tth time section.

[0102] Among them, the voltage square difference Q at the tth time section t and the nonlinear power flow equation f(P t ,Z) is as follows:

[0103]

[0104]

[0105] Where V it 、V jt Respectively represent the voltage of node i and node j at the tth time section. it , P jt are the active powers of nodes i and j at the tth time section respectively. it , Q jt are the reactive power of node i and node j at the tth time section respectively. it , R jt are the resistances of the zero-injection confluence node and node i line, and the zero-injection confluence node and node j line at the tth time section. it , X jt They are the reactances on the zero-injection confluence node and the line at node i, and the zero-injection confluence node and the line at node j at the tth time section respectively.

[0106] Embodiment 9:

[0107] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 8. Further, the conditions for determining the series-parallel relationship of the node pair (i, j) and the zero injection confluence node are as follows:

[0108] If the resistor |R j |>κ|R i |、Reactance|X j |>κ|X i | and resistance |R i |<τ、Reactance|X i |<τ, then node i is the parent node of node j, and there is no zero injection confluence node at this time, where κ and τ are both control parameters.

[0109] If the resistor |R i |>κ|R j |、Reactance|X i |>κ|X j | and resistance |R j |<τ、Reactance|X j |<τ, then node j is the parent node of node i, and there is no zero injection confluence node.

[0110] If neither of the above two conditions is met, then node i and node j are brother nodes, and the two nodes are in a parallel relationship. In this case, node i and node j are merged and regarded as a zero injection confluence node.

[0111] Embodiment 10:

[0112] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 1 to 9. Further, if node i is the parent node of node j, the updated voltage of the node to be determined is the voltage V of node i. i , the updated active power and reactive power of the node to be judged are as follows:

[0113]

[0114] Where V j is the voltage at node j, P i , P j are the active powers of nodes i and j respectively. i , Q j are the reactive power of node i and node j respectively. P' and Q' are the updated active power and reactive power of the node to be judged respectively. j is the line resistance at node j. j is the line reactance at node j.

[0115] If node j is the parent node of node i, the updated voltage of the node to be determined is the voltage V of node j. j , the updated active power and reactive power of the node to be judged are as follows:

[0116]

[0117] Where V i is the voltage at node i, R i is the line resistance at node i. i is the line reactance of node i.

[0118] If node i and node j are brother nodes, the updated voltage, active power, and reactive power of the node to be judged are as follows:

[0119]

[0120]

[0121] Where V k is the updated voltage of the node to be judged, V k(i) 、V k(j) are respectively the updated components of the voltage of the node to be determined at node i and node j.

[0122] Embodiment 11:

[0123] See also Figures 1 to 4 , a low voltage distribution network topology search method based on user meter data, comprising the following steps:

[0124] 1) Collect user meter data at m time sections and determine the number of sections for calculating the topology once, where m is a positive integer.

[0125] Since the calculation of node relationships requires solving nonlinear power flow equations, there must be a minimum number of sections that makes the solution stable. In this case, TS can be set to twice the minimum number of sections.

[0126] 2) Based on the number of sections calculated once and the total number of time sections m, K consecutive section samples TSS are randomly selected, and the sample ordinal number k=1, where K is a positive integer.

[0127] 3) Use the topology recognition algorithm to calculate the user recognition order IS belonging to the kth section sample TSS.

[0128] 4) Use the user identification sequence IS to identify the remaining K-1 section samples TSS in turn, obtain K-1 possible distribution network topologies, and store the K-1 possible distribution network topologies in the result set R.

[0129] 5) Determine whether k=K holds. If not, set k=k+1 and return to step 3). If so, proceed to step 6).

[0130] 6) The final distribution network topology is selected from the result set R by voting. There should be a total of K*(K-1) possible distribution network topologies in R. The K*(K-1) distribution network topology identification results are hard voted to finally obtain the most accurate distribution network topology structure.

[0131] Embodiment 12:

[0132] A low-voltage distribution network topology search method based on user meter data, the main technical content is shown in Example 11. Further, in order to ensure that there is overlapping data between every two TSSs, the value of the total number K of the section sample TSS is as follows:

[0133] K=a*(m÷TS) (1)

[0134] Where m is the total number of time sections. TS is the number of sections for calculating the topology once. a is the adjustment coefficient, and 1≤a≤2, which is determined based on the actual situation.

[0135] Embodiment 13:

[0136] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of Embodiments 11 to 12. Further, the step of using a topology recognition algorithm to calculate a user identification sequence IS belonging to the k-th section sample TSS comprises:

[0137] 3.1) Calculate the voltage distance matrix V of the node to be judged dist , and find the voltage distance matrix V dist The node pair (i, j) corresponding to the smallest non-zero value in .

[0138] In V dist Select the row and column corresponding to the smallest non-zero element, that is, the corresponding node i, j.

[0139] 3.2) Determine the series-parallel relationship of the node pair (i, j) and whether there is a zero injection sink node, and store the obtained series-parallel relationship of the node pair (i, j) into the series-parallel relationship set T. If there is a zero injection sink node, store the zero injection sink node into the zero injection power node set S.

[0140] 3.3) Based on the series-parallel relationship of the node pair (i, j) and the zero injection confluence node, update the voltage, active power, and reactive power of the node to be judged.

[0141] 3.4) Determine whether the number of updated nodes to be determined is 1. If so, output the zero injection power node set S and the series-parallel relationship set T of all nodes to obtain the user identification sequence IS. If not, return to step 3.1).

[0142] Embodiment 14:

[0143] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 11 to 13. Further, the voltage distance of the node to be judged is the Euclidean distance. Since there are n nodes to be judged, V dist is an n×n upper triangular matrix.

[0144] The voltage distance matrix V of the node to be judged dist The elements in Euclidean_distance are as follows:

[0145]

[0146] Where V i 、V j Respectively represent the voltage of node i and node j. t is the time section index, t = 1, 2, ..., m, m is the total number of time sections.

[0147] Embodiment 15:

[0148] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 11 to 14. Further, the step of determining the series-parallel relationship of the node pair (i, j) and whether there is a zero injection confluence node includes:

[0149] 3.2.1) Construct the linear overdetermined equation of the zero injection confluence node and solve it using the least squares method.

[0150] Since step 3.2.1) is the linearized model of the power flow formula between two nodes, the solved [R1 X1 R2 X2] T There is a certain error, which will affect the final judgment result, so a more accurate power flow model is needed to solve the line impedance for judgment.

[0151] 3.2.2) Based on the solution of the linear overdetermined equation, a nonlinear power flow equation is constructed, and the nonlinear power flow equation is solved iteratively through a numerical optimization algorithm (such as the Levenberg-Marquardt algorithm) to obtain the line impedance.

[0152] like Figure 4 As shown, this is a single line voltage drop model, and the nonlinear power flow equation can be obtained.

[0153] 3.2.3) Determine the series-parallel relationship of the node pair (i, j) and the zero injection confluence node based on the solved line impedance.

[0154] Embodiment 16:

[0155] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of Embodiments 11 to 15, further, as Figure 3 It is represented as a voltage drop diagram, where nodes 1 and 2 represent user nodes, whose voltage V, active power P, and reactive power Q are all known, and node 0 represents the zero injection confluence node that needs to be calculated, and the impedance on line 01 is Z1, and the impedance on line 02 is Z2, both of which are unknown. Thus, the linear power flow equation containing nodes 0, 1, and 2 can be obtained.

[0156] The linear overdetermined equation for the zero injection confluence node is as follows:

[0157]

[0158] In the formula, V0, V i 、V j are the voltages of zero injection into the confluence node, node i, and node j, respectively, P i , P j are the active powers of nodes i and j respectively. i , Q j are the reactive power of node i and node j respectively. i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line, respectively. i , X j They are the reactances on the zero-injection bus node and the node i line, and the zero-injection bus node and the node j line, respectively.

[0159] In order to eliminate the influence of the unknown voltage V0 at the zero-power node, the two equations in formula (3) can be subtracted to obtain:

[0160]

[0161] Embodiment 17:

[0162] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of Embodiments 11 to 16. Further, the calculation formula for solving the linear overdetermined equation using the least squares method is as follows:

[0163]

[0164] Where V it 、V jt Respectively represent the voltage of node i and node j in the tth time section. t=1,2...m, m is the total number of time sections. P it , P jt are the active powers of nodes i and j at the tth time section respectively. it , Qjt are the reactive power of node i and node j at the tth time section respectively. i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line, respectively. i , X j They are the reactances on the zero-injection bus node and the node i line, and the zero-injection bus node and the node j line, respectively.

[0165] Embodiment 18:

[0166] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of Embodiments 11 to 17, and further, the line impedance is as follows:

[0167]

[0168] e t (Z) = Q t -f(P t ,Z) (6)

[0169] Where t is the time section index, m is the total number of time sections, and RSS is the line impedance. t (Z) is the residual of the tth time section.

[0170] Among them, the voltage square difference Q at the tth time section t and the nonlinear power flow equation f(P t ,Z) is as follows:

[0171]

[0172]

[0173] Where V it 、V jt Respectively represent the voltage of node i and node j at the tth time section. it , P jt are the active powers of nodes i and j at the tth time section respectively. it , Q jt are the reactive power of node i and node j at the tth time section respectively. it , R jt are the resistances of the zero-injection confluence node and node i line, and the zero-injection confluence node and node j line at the tth time section. it , X jt They are the reactances on the zero-injection confluence node and the line at node i, and the zero-injection confluence node and the line at node j at the tth time section respectively.

[0174] Embodiment 19:

[0175] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 11 to 18. Further, the conditions for determining the series-parallel relationship of the node pair (i, j) and the zero injection confluence node are as follows:

[0176] If the resistor |R j |>κ|R i |、Reactance|X j |>κ|X i | and resistance |R i |<τ、Reactance|X i |<τ, then node i is the parent node of node j, and there is no zero injection confluence node at this time, where κ and τ are both control parameters.

[0177] If the resistor |R i |>κ|R j |、Reactance|X i |>κ|X j | and resistance |R j |<τ、Reactance|X j |<τ, then node j is the parent node of node i, and there is no zero injection confluence node.

[0178] If neither of the above two conditions is met, then node i and node j are brother nodes, and the two nodes are in a parallel relationship. In this case, node i and node j are merged and regarded as a zero injection confluence node.

[0179] Embodiment 20:

[0180] A low-voltage distribution network topology search method based on user meter data, the main technical content of which is shown in any one of embodiments 11 to 19. Further, if node i is the parent node of node j, i and j can be combined and regarded as node i, and the voltage of the node to be determined after the update is the voltage V of node i. i , the updated active power and reactive power of the node to be judged are as follows:

[0181]

[0182] Where V j is the voltage at node j, P i , P j are the active powers of nodes i and j respectively. i , Q j are the reactive power of node i and node j respectively. P' and Q' are the updated active power and reactive power of the node to be judged respectively. j is the line resistance at node j. j is the line reactance at node j.

[0183] If node j is the parent node of node i, i and j can be combined into node j, and the updated voltage of the node to be judged is the voltage V of node j. j , the updated active power and reactive power of the node to be judged are as follows:

[0184]

[0185] Where V i is the voltage at node i, R i is the line resistance at node i. i is the line reactance of node i.

[0186] If nodes i and j are sibling nodes, the zero injection confluence node is k (k is not any known actual user), i and j can be combined as node k, and the updated voltage, active power, and reactive power of the node to be judged are as follows:

[0187]

[0188]

[0189] Where V k is the updated voltage of the node to be judged, V k(i) 、V k(j) are respectively the updated components of the voltage of the node to be determined at node i and node j.

[0190] Embodiment 21:

[0191] See also Figures 1 to 4 , a low-voltage distribution network topology search method based on user meter data, the main technical contents are as follows:

[0192] The existing low voltage distribution network topology identification algorithm has a flow chart as follows: Figure 2 As shown, it has the following steps:

[0193] Step 1: Calculate the voltage distance matrix V of the node to be judged dist .

[0194] Step 2: At V dist Select the row and column corresponding to the smallest non-zero element, that is, the corresponding node i, j, and calculate the possibility of zero injection into the confluence node. The calculation formula is derived based on the nonlinear power flow equation between nodes.

[0195] Step 3: Update the voltage, active power and reactive power of the node to be judged. If the number of nodes to be judged is 1, all zero-power injection nodes in the network have been calculated. End. Otherwise, return to step 1. Specific output is performed through specified different modes.

[0196] In step 1, the Euclidean distance is used to calculate the voltage distance of the node to be judged, as shown in formula (1), where m represents the number of time sections, V i and V j Represents the voltage at node i and node j.

[0197]

[0198] Since there are n nodes to be judged, V dist is an n×n upper triangular matrix.

[0199] In step 2, the focus is on how to calculate the series-parallel connection relationship between the zero injection confluence node and the obtained node.

[0200] 1. Calculation of zero injection confluence nodes and determination of series-parallel relationships between nodes:

[0201] Step 1: Solving linear overdetermined equations

[0202] like Figure 3 It is represented as a voltage drop diagram, where nodes 1 and 2 represent user nodes, whose voltage V, active power P, and reactive power Q are all known, and node 0 represents the zero injection confluence node to be calculated, the impedance on line 01 is Z1, and the impedance on line 02 is Z2, both of which are unknown. Thus, the linear power flow equation including nodes 0, 1, and 2 can be obtained, as shown in formula (2).

[0203]

[0204] In order to eliminate the influence of the unknown voltage V0 of the zero-power node, the two equations in equation (2) can be subtracted to obtain equation (3).

[0205]

[0206] At this time, if the collected user meter data has m time sections, then formula (3) is rewritten into the form of an overdetermined equation (4), and the least squares method can be used to solve [R1 X1 R2 X2] T .

[0207]

[0208] Step 2: Solving nonlinear equations

[0209] Since formula (2) is a linearized model of the power flow formula between two nodes, the solved [R1 X1 R2 X2] T There is a certain error, which will affect the final judgment result, so a more accurate power flow model is needed to solve the line impedance for judgment. Figure 4 As shown, this is a single line voltage drop model, and the nonlinear power flow equation can be obtained as shown in (5).

[0210]

[0211] Similarly, subtracting the two equations in (5) eliminates We get formula (6).

[0212]

[0213] Under m time sections, it is written in matrix form as (7). Equation (7) is a nonlinear equation and can be solved iteratively using a numerical optimization algorithm (such as the Levenberg-Marquardt algorithm). The initial value can be selected using the impedance solved in step 1 as the initial value. The objective function is to minimize the residual sum of squares to solve the exact line impedance value Z = [R1, X1, R2, X2] T , the formula for the minimum parameter sum of squares is (8).

[0214]

[0215]

[0216] in represents the voltage square difference of the i-th time section. f(P i ,Z) is the nonlinear function of the tidal current at the i-th section.

[0217]

[0218] Step 3 Calculation of zero injection confluence nodes

[0219] The series-parallel relationship criterion of any observation node pair (i, j) (where i≠j) and the zero-injection confluence node determination principle are as follows:

[0220] 1) If condition 1 is met: |R j |>κ|R i |、|X j |>κ|X i | and|R i |<τ、|X i |<τ (where κ, τ are control parameters), then i is the parent node of j.

[0221] 2) If condition 2 is met: |Ri |>κ|R j |、|X i |>κ|X j | and|R j |<τ、|X j |<τ, then j is the parent node of i.

[0222] 3) If neither of the above two conditions is met, condition 3 is met, i and j are brother nodes, and the two nodes are in a parallel relationship.

[0223] In step 3, after determining whether there is a zero injection confluence node and the series-parallel relationship between nodes, the voltage, active power, and reactive power of the node to be determined need to be updated. After the update, if the number of nodes to be determined is equal to 1, the zero injection power node set S and the series-parallel relationship set T of all nodes are output.

[0224] 1) If condition 1 is met: there is no zero injection confluence node at this time, i and j can be combined as node i, and its voltage is V i , active power and reactive power are calculated as formula (10).

[0225]

[0226] 2) If condition 2 is met, there is no zero injection confluence node, i and j can be combined as node j, and its voltage V j , the active and reactive power are calculated as formula (11).

[0227]

[0228] 3) If condition 3 is met, the zero injection sink node is k (k is not any known actual user), i and j can be combined as node k, and the voltage calculation formula is as shown in (12), and the active and reactive power calculation is as shown in (13).

[0229]

[0230]

[0231] A low voltage distribution network topology search method based on user meter data

[0232] Based on the existing topology recognition algorithm, the present invention adds a distribution network topology search method for multiple section intervals. This method can reduce the accidental errors of the original single section interval data calculation algorithm and obtain a relatively accurate and stable topology structure. The flow chart is as follows: Figure 1 shown.

[0233] A multi-section search method for low-voltage distribution network topology based on user meter data includes the following steps, and its flow chart is as follows: Figure 1As shown, the top-down topology recognition algorithm flowchart is as follows Figure 2 shown.

[0234] Step 1: Determine the number of sections TS (time sections) required to calculate a topology.

[0235] Step 2: Based on the selected section number TS and the total number of existing data sections m, different K section samples TSS (Time section sample) are determined according to certain rules.

[0236] Step 3: Perform K-fold cross validation, selecting a cross-sectional sample TSS each time i As a test set, it is used to obtain the user identification sequence IS (Identification sequence). The remaining (K-1) are calculated according to the obtained user identification sequence based on the existing top-down topology recognition algorithm to calculate the corresponding topological structure. This is repeated K times to obtain a total of K*(K-1) topological search results.

[0237] Step 4: Voting to select the final topology.

[0238] The following will describe each step in detail:

[0239] In step 1, we first need to determine the number of sections TS required to calculate a node relationship. Since calculating the node relationship requires solving the nonlinear power flow equation, there must be a minimum number of sections that makes the solution stable. In this case, TS can be set to twice the minimum number of sections.

[0240] In step 2, assuming that the total number of data sections is m, K random but continuous time section samples TSS can be extracted. In order to ensure that there is overlapping data between every two TSS, the value of K is shown in formula (14). Where a is the adjustment coefficient, and the value range is 1≤a≤2, which is determined according to the actual situation.

[0241] K=a*(m÷TS) (14)

[0242] In step 3, after K-fold cross validation, only one TSS is selected each time, and the user identification sequence IS belonging to the TSS is obtained using mode 1 of the topology identification algorithm. This identification sequence is applied to mode 2, and the remaining (K-1) TSSs are identified in turn. The results are stored in R. At the end, R should have a total of K*(K-1) possible distribution network topologies.

[0243] Step 4: Hard voting is performed on the obtained K*(K-1) distribution network topology identification results to finally obtain the most accurate distribution network topology structure.

Claims

1. A low voltage distribution network topology search method based on user meter data, characterized in that: The following steps are involved: 1) Collect user meter data at m time sections and determine the number of sections for calculating the primary topology, where m is a positive integer; 2) Based on the number of sections calculated once and the total number of time sections m, randomly select K consecutive section samples TSS, and set the sample ordinal number k = 1, where K is a positive integer; 3) Use the topology recognition algorithm to calculate the user identification order IS belonging to the kth section sample TSS; 3.1) Calculate the voltage distance matrix V of the node to be judged dist , and find the voltage distance matrix V dist The node pair (i, j) corresponding to the minimum value that is not 0 in ; The voltage distance matrix V of the node to be judged dist The elements in Euclidean_distance are as follows: Where V it 、V jt Respectively represent the voltage of node i and node j; t is the time section index, t = 1, 2, ..., m, m is the total number of time sections; 3.2) Determine the series-parallel relationship of the node pair (i, j) and whether there is a zero injection sink node, and store the obtained series-parallel relationship of the node pair (i, j) into the series-parallel relationship set T. If there is a zero injection sink node, store the zero injection sink node into the zero injection power node set S; The step of determining the series-parallel relationship of the node pair (i, j) and whether there is a zero injection confluence node comprises: 3.2.1) Construct the linear overdetermined equation of the zero injection confluence node and solve it using the least squares method; The linear overdetermined equation for the zero injection confluence node is as follows: In the formula, V0, V i 、V j are the voltages of zero injection into the confluence node, node i, and node j, respectively, P i , P j are the active power of node i and node j respectively; Q i , Q j are the reactive power of node i and node j respectively; R i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line respectively; X i , X j are the reactances on the zero-injection bus node and the line at node i, and the zero-injection bus node and the line at node j, respectively; The calculation formula for solving the linear overdetermined equation using the least squares method is as follows: Where V it 、V jt Respectively represent the voltage of node i and node j at the tth time section; t=1,2...m, m is the total number of time sections; P it , P jt are the active power of node i and node j at the tth time section respectively; Q it , Q jt are the reactive power of node i and node j at the tth time section respectively; R i , R j are the resistances of the zero injection confluence node and node i line, and the zero injection confluence node and node j line respectively; X i , X j are the reactances on the zero-injection bus node and the line at node i, and the zero-injection bus node and the line at node j, respectively; 3.2.2) Based on the solution of the linear overdetermined equation, a nonlinear power flow equation is constructed, and the line impedance is obtained by solving the nonlinear power flow equation through a numerical optimization algorithm; The line impedance is as follows: e t (Z)=Q t -f(P t ,Z) (5) Where t is the time section index, m is the total number of time sections; RSS is the line impedance; e t (Z) is the residual of the tth time section; Among them, the voltage square difference Q at the tth time section t and the nonlinear power flow equation f(P t ,Z) is as follows: Where V it 、V jt Respectively represent the voltage of node i and node j at the tth time section; P it , P jt are the active power of node i and node j at the tth time section respectively; Q it , Q jt are the reactive power of node i and node j at the tth time section respectively; R it , R jt are the resistances of the zero-injection confluence node and the node i line, and the zero-injection confluence node and the node j line at the t-th time section; X it , X jt are the reactances of the zero-injection confluence node and the line at node i, and the zero-injection confluence node and the line at node j at the t-th time section, respectively; 3.2.3) Determine the series-parallel relationship of the node pair (i, j) and the zero injection confluence node based on the solved line impedance; The conditions for determining the series-parallel relationship of the node pair (i, j) and the zero injection confluence node are as follows: If the resistor |R j |>κR i |、Reactance|X j |>κX i | and resistance |R i |<τ、Reactance|X i |<τ, then node i is the parent node of node j, and there is no zero injection confluence node at this time, where κ and τ are both control parameters; If the resistor |R i |>κR j |、Reactance|X i |>κX j | and resistance |R j |<τ、Reactance|X j |<τ, then node j is the parent node of node i, and there is no zero injection confluence node; If neither of the above two conditions is met, then node i and node j are brother nodes, and the two nodes are in a parallel relationship. In this case, node i and node j are merged and regarded as a zero injection confluence node; 3.3) Based on the series-parallel relationship of the node pair (i, j) and the zero injection confluence node, update the voltage, active power, and reactive power of the node to be judged; 3.4) Determine whether the number of updated nodes to be determined is 1. If so, output the zero injection power node set S and the series-parallel relationship set T of all nodes to obtain the user identification sequence IS; if not, return to step 3.1); 4) Use the user identification sequence IS to identify the remaining K-1 section samples TSS in turn, obtain K-1 possible distribution network topologies, and store the K-1 possible distribution network topologies in the result set R; 5) Determine whether k=K is established. If not, set k=k+1 and return to step 3). If yes, proceed to step 6); 6) Select the final distribution network topology from the result set R by voting.

2. A low voltage distribution network topology search method based on user meter data according to claim 1, characterized in that: The value of the total number K of TSS of the cross-section samples is as follows: K=a*(m÷TS) (8) Where m is the total number of time sections; TS is the number of sections for calculating the topology once; a is the adjustment coefficient, and 1≤a≤2.

3. A low voltage distribution network topology search method based on user meter data according to claim 1, characterized in that: If node i is the parent node of node j, the updated voltage of the node to be determined is the voltage V of node i. i , the updated active power and reactive power of the node to be judged are as follows: Where V j is the voltage at node j, P i , P j are the active power of node i and node j respectively; Q i , Q j are the reactive power of node i and node j respectively; P' and Q' are the updated active power and reactive power of the node to be judged respectively; R j is the line resistance at node j; X j is the line reactance of node j; If node j is the parent node of node i, the updated voltage of the node to be determined is the voltage V of node j. j , the updated active power and reactive power of the node to be judged are as follows: Where V i is the voltage at node i, R i is the line resistance of node i; X i is the line reactance of node i; If node i and node j are brother nodes, the updated voltage, active power, and reactive power of the node to be judged are as follows: Where V k is the updated voltage of the node to be judged, V k(i) 、V k(j) are respectively the updated components of the voltage of the node to be determined at node i and node j.

Images