Power distribution network line fault positioning system based on multi-source information fusion

Abstract

The invention discloses a power distribution network line fault positioning system based on multi-source information fusion, which relates to the technical field of power system operation and maintenance and comprises a data acquisition module, a data preprocessing module, an electric network topology management module, a fault prediction module, a fault judgment module and a man-machine interaction and alarm module. The method comprises the following steps: acquiring running state information from a power distribution network in real time, generating a structured multi-dimensional information database according to the running state information acquired by a data acquisition module, and performing time alignment, format standardization and feature vector construction on the acquired running state information to generate the structured multi-dimensional information database; according to the method, a fault section is preliminarily judged according to fault features, a fault state prediction model is constructed, a multi-dimensional information database is used for estimating state variables of the model, and the method aims at overcoming the defects that in the prior art, fault positioning fuzziness and reliability are insufficient due to the fact that a single information source is relied on.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation and maintenance, and in particular to a distribution network line fault location system based on multi-source information fusion. Background Art

[0002] The safety, stability, and reliability of power systems are crucial. As the final link between the high-voltage transmission network and end users, the operating status of the distribution network is directly related to power quality and continuity. With socioeconomic development and the continued advancement of smart grid construction, distribution networks are becoming increasingly large and their topologies increasingly complex. The widespread integration of new elements such as distributed power sources, energy storage units, and electric vehicle charging stations has profoundly changed the operating modes and fault characteristics of traditional distribution networks.

[0003] Existing techniques compare the actual measured current waveform or its derived characteristics with pre-stored "fault fingerprints" in a database to infer the most likely fault point. However, as distribution networks evolve into highly dynamic and highly uncertain active complex networks, the inherent theoretical limitations and single-dimensional nature of fault location methods, particularly those relying solely on current measurement data, have become increasingly apparent. First, an extremely large and detailed fault sample library must be constructed; second, a large number of high-precision measurement devices must be added to the grid, significantly increasing system construction and maintenance costs. Summary of the Invention

[0004] The purpose of the present invention is to provide a distribution network line fault location system based on multi-source information fusion to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a distribution network line fault location system based on multi-source information fusion.

[0006] The system includes: data acquisition module, data preprocessing module, electrical network topology management module, fault prediction module, fault judgment module and human-computer interaction and alarm module.

[0007] Furthermore, the data acquisition module is used to obtain operating status information from the distribution network in real time; the data preprocessing module is used to time-align, format-standardize and construct feature vectors of the operating status information collected by the data acquisition module, and generate a structured multi-dimensional information database; the power network topology management module includes: a configuration information management unit and a directed graph model management unit; the fault section management module is used to make a preliminary judgment on the fault section based on the fault characteristics, and generate a set of candidate fault sections consisting of one or more fault line sections; the fault prediction module is used to construct a fault status prediction model for each candidate fault section, and use the multi-dimensional information database to estimate the state variables of the model and calculate the prediction vector of the fault status; the fault judgment module is used to determine the unique fault location and fault parameters according to the convergence conditions by performing several iterations on the evaluation status; the human-computer interaction and alarm module is used to visualize and publish the final fault location results.

[0008] Furthermore, the data acquisition module is communicatively connected to the data preprocessing module, the fault section management module is communicatively connected to the data preprocessing module and the electrical network topology management module, the fault prediction module is communicatively connected to the fault section management module, the data preprocessing module and the electrical network topology management module, the fault judgment module is communicatively connected to the fault prediction module, and the human-computer interaction and alarm module is communicatively connected to the fault judgment module.

[0009] Furthermore, the data acquisition module includes several intelligent terminal units and a data concentration unit; the intelligent terminal units are arranged at predetermined intervals along the distribution network line feeder, and when a fault occurs in the power grid, the intelligent terminal units collect traveling wave signals as fault signals; the data concentration unit is used to communicate with all intelligent terminal units.

[0010] Furthermore, the data preprocessing module includes: a time synchronization unit, a data standardization unit and a data management unit; the time synchronization unit is used to synchronize the timestamps contained in all data frames, construct a unified time axis, and align the data frames from different intelligent terminal units to the same time axis according to the timestamps; the data standardization unit is used to process the collected voltage data, and calculate the voltage sag vector composed of the voltage drop and phase angle offset at the measuring point of each intelligent terminal unit based on the voltage at the feeder head end. The voltage sag vector is a vector representation of a complex number composed of the voltage drop and phase angle offset; the data management unit is used to aggregate the preprocessed data to obtain a multidimensional information database.

[0011] Furthermore, the electrical network topology management module includes: a configuration information management unit and a directed graph model management unit; the configuration information management unit is used to manage the status information of electrical control devices in the power grid; the directed graph model management unit is used to manage the directed graph model in the power grid, and a directed graph model describing the power grid wiring method is constructed and dynamically maintained through a graph theory algorithm. In the directed graph model, the nodes represent transformers, busbars, switches or load points, and the edges represent line segments. The weights of the edges are defined by the physical parameters of the line segments corresponding to the edges. The physical parameters include the positive-sequence impedance, zero-sequence impedance or susceptance per unit length of the line segment.

[0012] Furthermore, the fault section management module includes: a first preliminary positioning unit and a second preliminary positioning unit; the first preliminary positioning unit is used to obtain the timestamps of the fault signal arriving at the two terminal detection units when any two terminal detection units detect the fault signal, which are respectively recorded as t i and t j , calculate the fault point to one of the two terminal detection units ITU i The distance d i , d i =(L ij + v ×(t i -t j )) / 2, where L ij represents the distance between the two terminal detection units, and v represents the propagation speed of the fault signal; the second preliminary positioning unit is used to collect voltage data and circuit data at the feeder outlet of the substation, and calculate the direction of transient energy flow within a preset period after the fault signal is detected. The transient energy flow represents the integral of the time-varying function of the electric power on the feeder over the preset period. When the integral result of the transient energy on a certain feeder is a positive number, it indicates that the fault has occurred on the certain feeder; the first preliminary positioning unit and the second preliminary positioning unit operate in parallel, and output a list including at least one fault line section. The fault line section in the list is used as a candidate section, and the starting node and the ending node of the fault line section constitute the unique identifier of the corresponding candidate fault section.

[0013] Furthermore, the fault prediction module includes: a fault state vector management unit, a fault state vector management unit and a nonlinear measurement unit; the fault state vector management unit: defines the fault state vector X, X=[d,R f ,C type ] T Where d is the distance between the fault point and the starting node of the section, R f is the fault transition resistance, C type is the fault type identifier; the initialization unit is used to obtain any candidate fault segment S k , initialize the candidate fault segment Sk The state vector X k,0 , X k,0 = [d k0 ,R f,0 , C type,0 ] T Among them, d k0 For S k The midpoint, R f,0 is the ratio of the initial fault current amplitude to the voltage drop, C type,0 To determine the initial value of the fault type, a state covariance matrix P0 is initialized, whose diagonal elements represent the uncertainty of the initial state estimate; the nonlinear measurement unit is used to define an actual measurement vector Z meas , which is composed as follows: Z meas =[V sag,1 , V sag,2 ,…,V sag,a ,…,V sag,N ,I fault,0 ,I branch,1 ,…, I branch,q ,…,I branch,M ] T , where V sag,a is the voltage sag vector measured at the ath intelligent terminal unit, I fault,0 is the total fault current phasor at the substation outlet, I branch,q is the current phasor flowing through the qth branch line. By performing the power system short-circuit calculation, the electrical quantity measurement values ​​that should be obtained at the measurement points of all intelligent terminal units in the entire network under the current fault state are predicted. Using the substation bus voltage as the power supply, a short-circuit power flow calculation program is run once to obtain the predicted measurement vector Z pred .

[0014] Furthermore, the fault judgment module includes: an iteration unit, a measurement residual calculation unit and a state update unit; the iteration unit is used to construct a set of 2n+1 Sigma points according to the rules of the unscented Kalman filter algorithm based on the u-1th evaluation state and the u-1th covariance, where n is the dimension of the state vector X, and propagate each Sigma point through the nonlinear measurement model to obtain a set of prediction vectors; based on the weighted mean and weighted covariance of the predicted measurement values, the predicted measurement vector Z of the u-1th iteration is calculated. pred,u-1 and the covariance matrix P of the actual measurement vector zz ;

[0015] The measurement residual calculation unit is used to calculate the cross-covariance matrix P between the state vector and the actual measurement vector xz , then, calculate the Kalman gain K u , Finally, the actual full-network synchronous measurement vector Z meas , calculate the measurement residual (Z meas - Z pred,u-1 ); The state update unit is used to update the state vector and covariance matrix X according to the following formula u ,

[0016] X u = X u-1 + K u ·(Z meas -Z pred,u ), where X u-1 Indicates the u-1th evaluation state, X u Indicates the evaluation status of the uth time; , where P u-1 represents the covariance matrix of the u-1th evaluation, P u Represents the covariance matrix of the u-th evaluation; perform several independent iterations on each candidate fault segment, and take the state vector corresponding to the candidate segment with the minimum measurement residual norm as the target state vector X final , obtain the final distance d of the fault point in the target state vector relative to the starting point of the fault line section final , the final transition resistance of the fault R f,final , and the final flag C of the fault type type,final .

[0017] Furthermore, the human-computer interaction and alarm module includes: a visualization management unit and an information prompt unit; the visualization management unit is used to mark the fault line section and fault point on the map of the geographic information system; the information prompt unit is used to send the location information of the fault line section and fault point to relevant management personnel of the power grid or the operation and maintenance management system of the power grid.

[0018] The system's workflow is as follows:

[0019] 1. All intelligent terminal units (ITUs) synchronously collect voltage and current waveforms at the time of the fault. ITUs located upstream of the fault (e.g., at the substation exit and switches A and B) detect the fault current, while downstream ITUs (switches C and D) do not. Simultaneously, all ITUs capture the fault transient traveling wave and record the arrival timestamp of each wave head. This data is then uploaded and aggregated to the data preprocessing module.

[0020] 2. The electrical network topology management module confirms that all current section switches are in the closed state and provides complete line impedance parameters.

[0021] 3. The fault segment management module is activated. The traveling wave ranging algorithm uses the arrival time difference between the ITU at the substation exit and the ITU at switch A to preliminarily calculate the fault location on the line segment between the substation and switch A. Simultaneously, the energy direction algorithm confirms that the fault energy flows out of the substation and does not pass through other branches. Therefore, the module generates a single candidate fault segment: S1 = [substation exit, switch A].

[0022] The fault section management module, whose input terminal is connected to the data preprocessing module and the electrical network topology management module, performs rapid preliminary analysis based on some highly deterministic fault characteristics, generating one or more candidate fault sections with high probability, thereby narrowing the search space for subsequent refined calculations. The module implements two preliminary location algorithms in parallel:

[0023] The first is a preliminary positioning algorithm based on traveling wave ranging. This algorithm starts when at least two intelligent terminal units detect the fault traveling wave signal. Assume that the two intelligent terminal units ITU that detect the initial traveling wave head are i and ITU j , the line distance between them is L ij (provided by the real-time network topology model), the time stamps of the traveling wave arriving at the two units are t i and t j , the propagation speed of the traveling wave in the line is v, then the distance from the fault point to the intelligent terminal unit ITU i The distance d i It can be calculated by the following formula: i =(L ij +v×(t i -t j )) / 2. This method can quickly identify the physical line segment where a fault point is located.

[0024] The second is a preliminary location algorithm based on head-end energy direction. This algorithm uses voltage and current data collected at the substation outlet to calculate the direction of transient energy flow during the initial transient phase of the fault (for example, within the first power frequency cycle after the fault occurs). This algorithm then determines whether the fault occurred in the feeder trunk or a specific branch, thereby eliminating non-faulty branches.

[0025] The fault section management module combines the results of the above two algorithms and outputs a list containing one or more candidate fault line sections, each of which is uniquely identified by its starting node and ending node.

[0026] 4. The fault prediction module starts the UKF estimator for the S1 section:

[0027] 4-1. Initialize the state vector X=[d,R f ,C type ]T According to the symmetrical component method, it is initially determined to be a phase B ground fault, that is, C type =BG; Estimate R based on the voltage and current amplitude of the first terminal f,initial =10 ohms; d initial =Length(S1) / 2.

[0028] 4-2. Engine builds network-wide synchronous measurement vector Z meas , which contains the fault current phasor at the substation outlet and the voltage sag vector composed of the voltage sag amplitude and phase angle measured synchronously at all ITUs (A, B, C, D, and end) along the line.

[0029] 4-3.UKF starts to iterate. In each iteration, according to the current [d, R f ] estimated value, set a 10 ohm B-phase ground fault at the corresponding position of the S1 section, and run the short-circuit calculation to predict the voltage sag values ​​of all measurement points such as A, B, C, D, and the end of the entire line, forming the predicted measurement vector Z pred .

[0030] 5. The fault diagnosis module makes a decision on the final fault location:

[0031] 5-1. By comparing Z pred and Z meas The measurement residual between is used to calculate the Kalman gain and to correct the state vector;

[0032] 5-2. After several iterations, the state vector X converges to a stable solution, where the measurement residual norm is minimized, and the judgment result is output.

[0033] Compared with existing technologies, the present invention offers the following advantages: The proposed distribution network line fault location system, based on multi-source information fusion, effectively integrates redundant information across the entire network through high-precision synchronous data acquisition and an advanced estimation algorithm centered on the UKF (United Kinetic Energy Function) framework. This allows for comprehensive identification of fault location, type, and parameters within a unified, precise physical model framework. This overcomes the inherent shortcomings of traditional methods in complex situations such as high-resistance grounding and dynamic topology changes, elevating the accuracy and reliability of fault location to a new level and providing strong technical support for rapid fault recovery and intelligent operation and maintenance of distribution networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a structural diagram of a cloud platform-based infrastructure project scheduling system of the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0036] Example: Figure 1 As shown, the present invention provides a technical solution, a cloud platform-based infrastructure project scheduling system.

[0037] The system includes: data acquisition module, data preprocessing module, electrical network topology management module, electrical network topology management module, fault prediction module, fault judgment module and human-computer interaction and alarm module.

[0038] Among them, the data acquisition module is communicated with the data preprocessing module, the fault section management module is communicated with the data preprocessing module and the electrical network topology management module, the fault prediction module is communicated with the fault section management module, the data preprocessing module and the electrical network topology management module, the fault judgment module is communicated with the fault prediction module, and the human-computer interaction and alarm module is communicated with the fault judgment module.

[0039] The data acquisition module is used to obtain real-time operating status information from the power distribution network;

[0040] Among them, the data acquisition module includes several intelligent terminal units and a data concentration unit; the intelligent terminal units are arranged at predetermined intervals along the distribution network line feeder. When a fault occurs in the power grid, the intelligent terminal units collect traveling wave signals as fault signals; the data concentration unit is used to communicate with all intelligent terminal units.

[0041] The data preprocessing module is used to perform time alignment, format standardization and feature vector construction on the operating status information collected by the data acquisition module to generate a structured multi-dimensional information database;

[0042] Among them, the data preprocessing module includes: a time synchronization unit, a data standardization unit and a data management unit; the time synchronization unit is used to synchronize the timestamps contained in all data frames, build a unified time axis, and align the data frames from different intelligent terminal units to the same time axis according to the timestamps; the data standardization unit is used to process the collected voltage data, and calculate the voltage sag vector composed of the voltage drop and phase angle offset at the measuring point of each intelligent terminal unit based on the voltage at the feeder head end. The voltage sag vector is a vector representation of a complex number composed of the voltage drop and phase angle offset; the data management unit is used to aggregate the preprocessed data to obtain a multidimensional information database.

[0043] The electrical network topology management module is used to manage the connection topology of electrical equipment in the power grid. The electrical network topology management module includes: a configuration information management unit and a directed graph model management unit; wherein the configuration information management unit is used to manage the status information of electrical control devices in the power grid;

[0044] The directed graph model management unit is used to manage the directed graph model in the power grid. It constructs and dynamically maintains a directed graph model that describes the grid wiring method through graph theory algorithms. In the directed graph model, nodes represent transformers, busbars, switches or load points, and edges represent line segments. The weight of the edge is defined by the physical parameters of the line segment corresponding to the edge. The physical parameters include the positive-sequence impedance, zero-sequence impedance or susceptance per unit length of the line segment.

[0045] The fault section management module is used to make a preliminary judgment on the fault section based on the fault characteristics and generate a candidate fault section set consisting of one or more faulty line sections;

[0046] The fault section management module includes: a first preliminary positioning unit and a second preliminary positioning unit; the first preliminary positioning unit is used to obtain the timestamps of the fault signal arriving at the two terminal detection units when any two terminal detection units detect the fault signal, which are respectively recorded as t i and t j , calculate the fault point to one of the two terminal detection units ITU i The distance d i , d i =(L ij + v ×(t i -t j )) / 2, where L ij represents the distance between the two terminal detection units, and v represents the propagation speed of the fault signal; the second preliminary positioning unit is used to collect voltage data and circuit data at the feeder outlet of the substation, and calculate the direction of transient energy flow within a preset period after the fault signal is detected. The transient energy flow represents the integral of the time-varying function of the electric power on the feeder over the preset period. When the integral result of the transient energy on a certain feeder is a positive number, it indicates that the fault has occurred on the certain feeder; the first preliminary positioning unit and the second preliminary positioning unit operate in parallel, and output a list including at least one fault line section. The fault line section in the list is used as a candidate section, and the starting node and the ending node of the fault line section constitute the unique identifier of the corresponding candidate fault section.

[0047] The fault prediction module is used to build a fault state prediction model for each candidate fault section, estimate the state variables of the model using the multi-dimensional information database, and calculate the prediction vector of the fault state;

[0048] Among them, the fault prediction module includes: a fault state vector management unit, a fault state vector management unit and a nonlinear measurement unit; the fault state vector management unit: defines the fault state vector X, X=[d,R f ,C type ] T Where d is the distance between the fault point and the starting node of the section, R f is the fault transition resistance, C type is the fault type identifier; the initialization unit is used to obtain any candidate fault segment S k , initialize the candidate fault segment S k The state vector X k,0 , X k,0 = [d k0 , R f,0 ,C type,0 ] T Among them, d k0 For S k The midpoint, R f,0 is the ratio of the initial fault current amplitude to the voltage drop, C type,0 To determine the initial value of the fault type, a state covariance matrix P0 is initialized, whose diagonal elements represent the uncertainty of the initial state estimate; the nonlinear measurement unit is used to define an actual measurement vector Z meas , which is composed as follows: Z meas =[V sag,1 , V sag,2 ,…,V sag,a ,…,V sag,N ,I fault,0 ,I branch,1 ,…, I branch,q ,…,I branch,M ] T , where V sag,a is the voltage sag vector measured at the ath intelligent terminal unit, I fault,0 is the total fault current phasor at the substation outlet, I branch,q is the current phasor flowing through the qth branch line. By performing the power system short-circuit calculation, the electrical quantity measurement values ​​that should be obtained at the measurement points of all intelligent terminal units in the entire network under the current fault state are predicted. Using the substation bus voltage as the power supply, a short-circuit power flow calculation program is run once to obtain the predicted measurement vector Z pred .

[0049] The fault diagnosis module is used to determine the unique fault location and fault parameters by iterating the evaluation status several times according to the convergence conditions; the human-computer interaction and alarm module is used to visualize and publish the final fault location results;

[0050] Among them, the fault judgment module includes: an iteration unit, a measurement residual calculation unit and a state update unit; the iteration unit is used to construct a set of 2n+1 Sigma points according to the rules of the unscented Kalman filter algorithm, based on the u-1th evaluation state and the u-1th covariance, each Sigma point carries a set of mean weights and covariance weights, where n is the dimension of the state vector X, and each Sigma point is propagated through the nonlinear measurement model to obtain a set of prediction vectors; based on the weighted mean and weighted covariance of the predicted measurement values, the predicted measurement vector Z of the u-1th iteration is calculated. pred,u-1 and the covariance matrix P of the actual measurement vector zz ;

[0051] The measurement residual calculation unit is used to calculate the cross-covariance matrix P between the state vector and the actual measurement vector xz , then, calculate the Kalman gain K u , Finally, the actual full-network synchronous measurement vector Z meas , calculate the measurement residual (Z meas - Z pred,u-1 ); The state update unit is used to update the state vector and covariance matrix X according to the following formula u ,

[0052] X u = X u-1 + K u ·(Z meas -Z pred,u ), where X u-1 Indicates the u-1th evaluation state, X u Indicates the evaluation status of the uth time; , where P u-1 represents the covariance matrix of the u-1th evaluation, P u Represents the covariance matrix of the u-th evaluation; perform several independent iterations on each candidate fault segment, and take the state vector corresponding to the candidate segment with the minimum measurement residual norm as the target state vector X final , obtain the final distance d of the fault point in the target state vector relative to the starting point of the fault line section final , the final transition resistance of the fault R f,final , and the final flag C of the fault type type,final .

[0053] The human-computer interaction and alarm module is used to visualize and publish the final fault location results, wherein the human-computer interaction and alarm module includes: a visualization management unit and an information prompt unit; the visualization management unit is used to mark the fault line section and fault point on the map of the geographic information system; the information prompt unit is used to send the location information of the fault line section and fault point to the relevant management personnel of the power grid or the power grid operation and maintenance management system.

[0054] In the examples:

[0055] A 10kV overhead distribution feeder F1 has a total length of 18.5 kilometers. The feeder adopts a radial structure. After it is drawn from substation S, three section switches A, B, and C are connected in series along the line, naturally dividing the line into four main sections: SA (4.2 km), AB (5.5 km), BC (6.1 km), and CD (2.7 km) from C to the line end D.

[0056] According to the deployment scheme of the present invention, eight sets of intelligent terminal units described in detail above were deployed at the exit of substation S, on both sides of section switches A, B, and C, and at the line end D. At 14:32:15 on a certain day, an unknown fault occurred on the line.

[0057] The system works according to the following process:

[0058] 1. Data acquisition and preprocessing: At the moment of the fault, all eight ITUs along the line synchronously collected the voltage and current waveforms at the time of the fault, with a sampling rate of 2.56MHz. The ITUs upstream of the fault point (substation outlet S, both sides of switch A, and the outlet side of switch B) all detected significant fault currents, while the ITUs downstream (both sides of switch C and terminal D) detected no fault current or very low currents.

[0059] At the same time, the ITUs at S, A, and B all capture the fault transient traveling waves and record their respective wave head arrival timestamps through synchronized clocks.

[0060] The specific record is: S =4:32:15.123456789 (seconds), t A =14:32:15.123470889 (seconds), t B =14:32:15.123489389 (seconds), all data collected by the ITU are uploaded to the data preprocessing module for time alignment, and a voltage sag vector containing the synchronized voltage sag values ​​of the eight measurement points along the entire line is generated.

[0061] 2. Topology modeling: The electrical network topology management module communicates with the distribution management system to confirm that the current segment switches A, B, and C are all in the closed state and the network is a complete chain structure, and obtains the precise line impedance parameters of each section.

[0062] 3. Preliminary fault diagnosis: The fault section management module is activated. First, the traveling wave ranging algorithm is activated. The ITU at points S and A is used for calculation. The line distance between the two points LSA = 4200 meters, and the traveling wave arrival time difference Δt = t A -t S =14.1 microseconds;

[0063] Assume that the speed of the traveling wave v = 299.7 m / μs. Calculate the distance d from the fault point to S. S =(L SA +v×(t S -t A )) / 2=(4200+299.7×(-14.1)) / 2≈-11.4 meters. The physical meaning of this result is unclear. Replace it with the ITU at points A and B for calculation. The line distance L between the two points AB =5500 meters, time difference Δt=t B -t A = 18.5 microseconds. Calculate the distance d from the fault point to A A =(L AB +v×(t A -t B )) / 2 = (5500 + 299.7 × (-18.5)) / 2 ≈ -21.2 meters, which also has unclear physical meaning. At this point, the algorithm automatically adopts single-ended ranging mode, using reflected waves. The time difference between the initial wave and the reflected wave from the fault point detected at S is 25.2 microseconds, and the fault distance is calculated as d = v × Δt / 2 = 299.7 × 25.2 / 2 = 3776.2 meters. This result points to the SA segment. Simultaneously, the head-end energy direction algorithm confirms that the fault energy is flowing out of substation S. Based on this comprehensive assessment, the module generates a unique candidate fault segment: S1 = [substation exit S, switch A], with a length of 4.2 km.

[0064] 4. Iterative state estimation: The fault prediction module starts the UKF estimator for the S1 segment.

[0065] 4-1. Initialize the state vector X=[d,R f ,C type ] T According to the asymmetry of the three-phase current measured at S, it is preliminarily determined that the C phase is grounding fault, that is, C type =CG; Estimate R based on the voltage and current amplitude at the first end f,initial =80 ohms, which is high resistance grounding; d initial=3776.2 meters (initial positioning result using traveling waves).

[0066] 4-2. Engine builds network-wide synchronous measurement vector Z meas , which contains the voltage sag vector composed of the voltage sag amplitude and phase angle measured synchronously at all ITUs such as S, A, B, C, and D, as well as the fault current phasor at S.

[0067] 4-3. UKF starts to iterate. In each iteration, the engine calculates the value of [d, R f ] estimated value, at the corresponding position of the SA section, for example, at the initial position 3776.2 meters, set an 80 ohm C phase grounding fault, at this time the transition resistance R f =80 ohms, fault type is C type =CG, and run the short-circuit calculation. First, obtain the original admittance matrix Y of each node in the power grid. Then, add a virtual node k in the SA section at a distance of 3776.2 meters from the substation, divide SA into two sections, Sk and kA, and calculate the newly added mutual admittance Y for each section. Sk and Y kA ;

[0068] Add an admittance branch y between phase C of the virtual node k and the ground f =1 / R f , calculate the C-phase self-admittance Y of virtual node k kk , Y kk =(y Sk +y kA )+y f , where y Sk represents the admittance between S and virtual node k, y kA Represents the admittance between virtual node k and A, and Y Sk 、Y kA and Y kk Add to the original admittance matrix Y to get the updated admittance matrix Y bus , obtain the voltage vectors of 8 measurement points to form vector V, and establish the node voltage equation I=Y bus V;

[0069] Compare the calculated results with the voltage of the node during normal operation to obtain the voltage sag values ​​of all 8 measurement points on the entire line, and arrange them in order of nodes to form the predicted measurement vector Z pred .

[0070] 4-4. By comparing Z pred and Z meas The measurement residual between them is used to calculate the Kalman gain and to correct the state vector [d,R f],After the first iteration, it is found that the voltage sag amplitudes of all actual measurement points are generally smaller than the predicted values, which indicates that the actual transition resistance may be larger than 80 ohms. At the same time, the voltage drop at S is slightly larger than the predicted value, while the voltage drop at A is slightly smaller than the predicted value, which indicates that the fault point may be closer to S than 3776.2 meters. Based on these residuals, the UKF algorithm automatically adjusts the estimated value of Rf upward and adjusts the estimated value of d toward S.

[0071] 4-5. After 6 iterations, the state vector X converges to a stable optimal solution, the result is: d final = 3652 meters, that is, 3652 meters from the substation S exit, R f,final =115.3 ohms, C type =CG, the measurement residual norm at this time reaches the global minimum, indicating that the fault prediction location of the entire network calculated based on this parameter on the physical model.

[0072] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A distribution network line fault location system based on multi-source information fusion, characterized by: The system includes: Data acquisition module, data preprocessing module, electrical network topology management module, fault prediction module, fault judgment module and human-computer interaction and alarm module; The data acquisition module is used to obtain operating status information from the power distribution network in real time; The data preprocessing module is used to perform time alignment, format standardization and feature vector construction on the operating status information collected by the data collection module to generate a structured multi-dimensional information database; The electrical network topology management module is used to manage the connection topology of electrical equipment in the power grid; The data preprocessing module is used to perform time alignment, format standardization and feature vector construction on the operating status information collected by the data collection module to generate a structured multi-dimensional information database; The fault section management module is used to make a preliminary judgment on the fault section according to the fault characteristics and generate a candidate fault section set consisting of one or more faulty line sections; The fault prediction module is used to construct a fault state prediction model for each candidate fault section, and use the multi-dimensional information database to estimate the state variables of the model and calculate the prediction vector of the fault state; The fault judgment module is used to determine a unique fault location and fault parameters according to a convergence condition by performing several iterations on the evaluation state; The human-computer interaction and alarm module is used to visualize and publish the final fault location results.

2. A distribution network line fault location system based on multi-source information fusion according to claim 1, characterized in that: The data acquisition module includes several intelligent terminal units and data concentration units; The intelligent terminal units are arranged at predetermined intervals along the feeder lines of the power distribution network, and when a fault occurs in the power grid, the intelligent terminal units collect traveling wave signals as fault signals; The data concentration unit is used to establish communication connections with all intelligent terminal units.

3. A distribution network line fault location system based on multi-source information fusion according to claim 2, characterized in that: The intelligent terminal unit includes a three-phase voltage transformer interface, a three-phase current transformer interface, an analog-to-digital conversion subunit, a clock synchronization subunit and a communication subunit.

4. The distribution network line fault location system based on multi-source information fusion according to claim 2, characterized in that: The data preprocessing module includes: time synchronization unit, data standardization unit and data management unit; The time synchronization unit is used to synchronize the timestamps contained in all data frames, build a unified time axis, and align the data frames from different intelligent terminal units to the same time axis according to the timestamps; The data normalization unit is used to process the collected voltage data and calculate the voltage sag vector composed of the voltage drop amplitude and phase angle offset at the measurement point of each intelligent terminal unit based on the voltage at the feeder head end. The voltage sag vector is a vector representation of the complex number composed of the voltage drop amplitude and phase angle offset. The data management unit is used to collect the pre-processed data to obtain a multi-dimensional information database.

5. The distribution network line fault location system based on multi-source information fusion according to claim 4, characterized in that: The electrical network topology management module includes: a configuration information management unit and a directed graph model management unit; The configuration information management unit is used to manage the status information of electrical control devices in the power grid; The directed graph model management unit is used to manage the directed graph model in the power grid. It constructs and dynamically maintains a directed graph model that describes the power grid wiring method through a graph theory algorithm. In the directed graph model, nodes represent transformers, busbars, switches or load points, and edges represent line segments. The weight of the edge is defined by the physical parameters of the line segment corresponding to the edge. The physical parameters include the positive-sequence impedance, zero-sequence impedance or susceptance per unit length of the line segment.

6. A distribution network line fault location system based on multi-source information fusion according to claim 5, characterized in that: The fault section management module includes: a first preliminary positioning unit and a second preliminary positioning unit; The first preliminary positioning unit is used to obtain the timestamps of the fault signals arriving at the two terminal detection units when any two terminal detection units detect the fault signals, which are respectively recorded as t i and t j , calculate the fault point to one of the two terminal detection units ITU i The distance d i , d i =(L ij + v ×(t i -t j )) / 2, where L ij represents the distance between the two terminal detection units, and v represents the propagation speed of the fault signal; The second preliminary positioning unit is used to collect voltage data and circuit data at the feeder outlet of the substation, and calculate the direction of transient energy flow within a preset period after the fault signal is detected. The transient energy flow represents the integral of the time-varying function of the electric power on the feeder over the preset period. When the integral result of the transient energy on a certain feeder is a positive number, it indicates that a fault has occurred on the feeder; The first preliminary positioning unit and the second preliminary positioning unit run in parallel, outputting a list including at least one faulty line section, taking the faulty line section in the list as a candidate section, and forming a unique identifier of the corresponding candidate faulty section by the starting node and the ending node of the faulty line section.

7. The distribution network line fault location system based on multi-source information fusion according to claim 6, characterized in that: The fault prediction module includes: a fault state vector management unit, a fault state vector management unit and a nonlinear measurement unit; Fault state vector management unit: define fault state vector X, X=[d,R f ,C type ] T Where d is the distance between the fault point and the starting node of the section, R f is the fault transition resistance, C type Fault type identification; The initialization unit is used to obtain any candidate fault segment S k , initialize the candidate fault segment S k The state vector X k,0 , X k,0 = [d k0 , R f,0 , C type,0 ] T Among them, d k0 For S k The midpoint, R f,0 is the ratio of the initial fault current amplitude to the voltage drop, C type,0 In order to determine the initial value of the fault type, a state covariance matrix P0 is initialized at the same time, whose diagonal elements represent the uncertainty of the initial state estimate; The nonlinear measurement unit is used to define an actual measurement vector Z meas , which is composed as follows: Z meas =[V sag,1 ,V sag,2 ,…,V sag,a ,…,V sag,N ,I fault,0 ,I branch,1 ,…, I branch,q ,…,I branch,M ] T , where V sag,a is the voltage sag vector measured at the ath intelligent terminal unit, I fault,0 is the total fault current phasor at the substation outlet, I branch,q is the current phasor flowing through the qth branch line. By performing the power system short-circuit calculation, the electrical quantity measurement values ​​that should be obtained at the measurement points of all intelligent terminal units in the entire network under the current fault state are predicted. Using the substation bus voltage as the power supply, a short-circuit power flow calculation program is run once to obtain the predicted measurement vector Z pred .

8. The distribution network line fault location system based on multi-source information fusion according to claim 7, characterized in that: The fault judgment module includes: an iteration unit, a measurement residual calculation unit and a state update unit; The iterative unit is used to construct a set of 2n+1 Sigma points according to the rules of the unscented Kalman filter algorithm based on the u-1th evaluation state and the u-1th covariance, where n is the dimension of the state vector X, and propagate each Sigma point through the nonlinear measurement model to obtain a set of prediction vectors; According to the weighted mean and weighted covariance of the predicted measurement values, the predicted measurement vector Z of the u-1 iteration is calculated pred,u-1 and the covariance matrix P of the actual measurement vector zz ; The measurement residual calculation unit is used to calculate the cross-covariance matrix P between the state vector and the actual measurement vector xz , then, calculate the Kalman gain K u , Finally, the actual full-network synchronous measurement vector Z meas , calculate the measurement residual (Z meas - Z pred,u-1 ); The state update unit is used to update the state vector and covariance matrix X according to the following formula u , X u = X u-1 + K u ·(Z meas -Z pred,u ), where X u-1 Indicates the u-1th evaluation state, X u Indicates the evaluation status of the uth time; , where P u-1 represents the covariance matrix of the u-1th evaluation, P u represents the covariance matrix of the u-th evaluation; Perform several independent iterations for each candidate fault segment, and take the state vector corresponding to the candidate segment with the minimum measurement residual norm as the target state vector X final , obtain the final distance d of the fault point in the target state vector relative to the starting point of the fault line section final , the final transition resistance of the fault R f,final , and the final flag C of the fault type type,final .

9. The distribution network line fault location system based on multi-source information fusion according to claim 8, characterized in that: The human-computer interaction and alarm module includes: visual management unit and information prompt unit; The visualization management unit is used to mark the fault line section and fault point on the map of the geographic information system; The information prompt unit is used to send the location information of the fault line section and the fault point to the relevant management personnel of the power grid or the operation and maintenance management system of the power grid.

10. The distribution network line fault location system based on multi-source information fusion according to claim 1, characterized in that: The data acquisition module is communicated with the data preprocessing module, the fault section management module is communicated with the data preprocessing module and the electrical network topology management module, the fault prediction module is communicated with the fault section management module, the data preprocessing module and the electrical network topology management module, the fault judgment module is communicated with the fault prediction module, and the human-computer interaction and alarm module is communicated with the fault judgment module.

Images